Experimental Studies of Seed-Grinding in Traditional Aboriginal Australia

Pieces of a ground stone puzzle: Experimental studies of seed-grinding in traditional Aboriginal Australia.

John Mildwaters Bachelor of Arts Bachelor of Social Science (Honours)

A thesis submitted for the degree of Master of Philosophy at The University of Queensland in 2020 School of Social Science

Table of Contents

Abstract ...... iv Declaration by author ...... vi Publications during candidature ...... vii Publications included in this thesis ...... vii Contributions by others to the thesis ...... ix Statement of parts of the thesis submitted to qualify for the award of another degree ...... ix Research Involving Human or Animal Subjects ...... ix Acknowledgements ...... x Financial support ...... x Keywords ...... xi Australian and New Zealand Standard Research Classifications (ANZSRC) ...... xi Fields of Research (FoR) Classification ...... xi ORCID ...... xi List of Figures ...... xii List of Tables ...... xiii 1 Chapter 1 - Introduction ...... 1 1.1 The Thesis ...... 1 1.1.1 Overarching Aims ...... 1 1.1.2 Specific Aims ...... 2 1.2 Scope and Significance ...... 2 1.3 Background ...... 3 1.4 Debates in Australian Archaeology ...... 5 1.5 Global Directions in Grindstone Research ...... 9 1.6 Methods ...... 16 1.7 Seed-Grinding in Theoretical Context ...... 16 1.8 Experimental Archaeology ...... 19 1.9 Introductory Conclusion ...... 23 1.10 The Papers ...... 23 1.11 References ...... 24 2 Chapter 2 - Seed Grinding in Traditional Aboriginal Australia ...... 25 2.1 Introduction ...... 25 2.2 Methodology ...... 25

2.3 Conclusion in summary ...... 25 2.4 The Paper ...... 25 3 Chapter 3 - Productive Efficiency of Various Grindstones ...... 45 3.1 Introduction ...... 45 3.2 Methodology ...... 45 3.3 Conclusion in Summary ...... 45 3.4 The Second Paper ...... 45 4 Chapter 4 - Productivity of Native Seeds when Ground ...... 73 4.1 Introduction ...... 73 4.2 Methodology ...... 73 4.3 Conclusion in Summary ...... 73 4.4 The Third Paper ...... 73 5 Chapter 5 - Summary and Conclusions ...... 111 5.1 Summary ...... 111 5.2 Concluding Remarks ...... 118 5.3 References ...... 120

iii

Abstract

The core of this thesis consists of three research papers dealing with various aspects of Australian Aboriginal seed-grinding— largely from an experimental perspective. The first paper uses ethnographic and excavation reports to review and assess the seed-grinding literature and evaluate the role of grindstones in traditional seed-based Australian economies. It highlights debates in Australian archaeology involving seed-grinding and grindstones and isolates possible methodological problems revolving around grinding issues including the relative productive efficiency of the grindstone and seed combination being used. The baseline technological factors that affect flour or meal production are shown to be poorly understood and the paper suggests aspects suitable for further investigation via experiment.

The second paper presents a controlled experimental examination of the efficiency of a range of replicated sandstone grindstones with a variety of surface morphologies based on their proficiency in processing grain into meal or flour. The results show that large millstones significantly outperform smaller grindstones but that other factors can also affect grinding performance. Three proxy domesticated grains were utilised and variations in the grinding performance of the grains demonstrated the need for a full evaluation of the grinding profiles of traditionally used native seeds.

The final paper assesses the grinding characteristics of several native seeds, in both raw and pre-treated forms, ground by Aboriginal Australians. The soft seeds of grasses and herbs (grass-type seeds) were found to have comparable grinding characteristics and grind in largely similar times without need for pre-conditioning. However, hard seeds from trees and shrubs often benefited from parching or soaking before being ground. Unlike the soft seeds, the grinding times for hard seeds varied widely and no generalisations about the grinding characteristics of these seeds could be proffered— each seed needs to be assessed on an individual basis.

The problems identified and investigated in the papers illustrate that, in general, only a broad grained understanding of seed-grinding is available. Despite the inclusion of seed- grinding data in influential substantive and theoretical works, the baseline information on which those works rely, when removed from their original contexts, can be incomplete or inappropriate and thus possibly misleading.

The experiments conducted clarify a number of uncertainties concerning seed-grinding whilst highlighting the many remaining technological aspects of grinding implements, and the seeds they ground, which requires further urgent ethnoarchaeological and experimental investigation.

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

Publications during candidature Peer-reviewed papers:

Mildwaters, John. 2018. Seed-grinding stones: a review from a mainly Australian perspective. The Artefact 2016. Vol. 39. pp. 30-41.

Mildwaters, John and Clarkson, Chris. 2018. The efficiency of Australian grindstones for processing seed: a quantitative experiment using reproduction implements and controlling for morphometric variation and grinding techniques. Journal of Archaeological Science: Reports. Volume 17. pp. 7-18.

Mildwaters, John and Clarkson, Chris. 2020. An experimental assessment of the grinding characteristics of some native seeds used by Aboriginal Australians. Journal of Archaeological Science: Reports. Volume 30. Article 102127. pp. 1-15

Publications included in this thesis Peer-reviewed papers:

Incorporated in Chapter 2:

Mildwaters, John. 2018. Seed-grinding stones: a review from a mainly Australian perspective. The Artefact 2016. Vol. 39. pp. 30-41.

Incorporated in Chapter 3:

vii

Contributor Statement of contribution Author Mildwaters (Candidate) Conception and design (80%) Analysis and interpretation (80%) Drafting and production (80%) Author Clarkson Conception and design (20%) Analysis and interpretation (20%) Drafting and production (20%)

Incorporated in Chapter 4:

viii

Contributions by others to the thesis

No contributions by others.

Statement of parts of the thesis submitted to qualify for the award of another degree

None.

Research Involving Human or Animal Subjects

No animal or human subjects were involved in this research.

Acknowledgements

This study was undertaken late in life and in small snippets of time when business and family concerns allowed. Its completion would not have been possible without the patience and help of many people, only a few of whom can be acknowledged here. University staff in various areas gave unstinting support to my somewhat unusual circumstances. I wish to sincerely thank my supervisor Professor Chris Clarkson, as well as my associate supervisor Dr Tiina Manne and milestone assessors, Associate Professor Andy Fairbairn and Dr Alison Crowther, for awakening, and enthusiastically supporting, my interest in Aboriginal seed-grinding technology. The inputs of reviewers of the published papers and of the thesis overall were extremely valuable and these anonymous persons are gratefully acknowledged.

I also wish to thank my somewhat bemused, but unstintingly supportive, family for their patience and assistance as the project consumed virtually my every spare moment. Thanks to my wife Heather for proof reading (and punctuation lessons), and children Brett and Lisa for regular IT support, in particular Brett, who managed to arrange my two-finger word processing into documents acceptable for publication.

Financial support

No financial support was provided to fund this research.

Keywords grindstones, morphology, productivity, seed-grinding, seeds, Australia, Aboriginal.

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 210101, Aboriginal and Torres Strait Islander Archaeology, 60% ANZSRC code: 210102, Archaeological Science, 40%

Fields of Research (FoR) Classification

FoR code:

2101 Archaeology Humanities and Creative Arts 50%

2199 Other History and Archaeology Humanities and Creative Arts 50%

ORCID Registration Number: 0000-0001-9943-4236

List of Figures

List of Figures for Chapters 2 to 4 appear within the publications.

xii

List of Tables

Chapter 1 1. A sample of global grinding research (excluding papers based predominantly on use-wear and residue analysis).

List of Tables for Chapters 2 to 4 appear within the publications.

xiii

1 Chapter 1 - Introduction

Grinding tools of a variety of models have been manufactured and used by hunters and gatherers, horticulturalists and agriculturalists. Critical studies related to particular and contextual differences in tool design and tool use among cultural groups, is … of critical importance (Lidstrom Holmberg 1998:133).

1.1 The Thesis

1.1.1 Overarching Aims Seed-grinding was a core component of many Aboriginal economies in Australia for the last 65 ka, particularly in arid areas where people at times depended on seeds for survival (Clarkson et al. 2017). The material remains of seed-grinding have therefore underpinned many archaeological propositions about economic, technological, societal and demographic change in Australia since its peopling (e.g. Bowdler 1977; Cane 1984; Lourandos 1997; Smith 2013). In addition, seed-grinding data have featured in high-level models of hunter-gatherer economics and ecology, both in Australia and elsewhere (e.g. Bright et al. 2002; O’Connell and Hawkes 1981).

The implications of the economic use of plants in traditional Aboriginal diets impact widely on aspects of concern to archaeology in Australia. In particular, the possible dietary contribution of ground seeds has generated significant discussion (e.g. Gorecki et al. 1997; Hayes 2015; Smith 1985; 1986; 1989; 2015; Veth and O’Connor 1996). Yet surprisingly little is known about the base-line issues which determined the acquisition and use of the grinding implements or the selection and treatment of the seeds they processed. Each species of seed incorporated into a diet required a chain of decisions by responsible actors intent on furthering their own lives (Bird and O’Connell 2006). Was the seed edible and nutritious? Did it grind readily? What additional processing was required? Were current toolkits suitable for the task? Or would additional technology need to be acquired? At what cost? Were available labour hours sufficient? Should the seed be rejected in favour of other economic alternatives? And so on. Such fine-grained details are critical to any attempt to understand traditional lifeways in which seed use was of economic importance. However, at present, little detailed, transposable information is available.

It is thus the primary aim of this thesis to highlight and examine deficiencies in our understanding of Australian Aboriginal seed-grinding and to provide low-level quantitative data to help address these issues.

1.1.2 Specific Aims This thesis (including the three research papers forming its core) will- 1. Review debates and archaeological evidence for seed-grinding with a particular focus on the types of resources used, the technologies involved and the economics of seed-grinding in traditional Australian economies; 2. Explore the efficiency of a range of traditional grindstones known from ethnographic and archaeological contexts; and 3. Investigate the grinding profiles of various native seeds known from ethnographic reports to have been incorporated into customary diets.

To accomplish each of these research aims, this thesis will review known ethnographic information on seed-grinding tools, methods of preparation and processing of a wide range of seeds and perform novel experiments with replicated grindstones and both commercial proxy seeds and native Australian seeds known to have been ground in the past.

1.2 Scope and Significance

The grindstones replicated and examined can be termed ‘central Australia’ types, excepting a mortar, which was widely distributed, and a basin millstone more usually associated with the contingent dry areas of south east Australia (Beveridge 1889; Smith 1985). Although these grindstones are representative of the most common types, several morphologically different grindstones were used in other areas and these need to be subjected to detailed research (Smith 2015). Similarly, only a small proportion of the seeds traditionally used by Aboriginal Australians are evaluated due to limited availability.

The study is significant in that it delivers the only comprehensive and critical appraisals of, firstly, the comparative efficiency of a diverse range of grindstones and, secondly, the productivity of seeds from plants ranging from herbs to grasses to shrubs and trees. Also

provided is a preliminary, but not previously reported, evaluation of the use of domesticated seeds as grinding analogues for Australian native seeds.

The productive potential of a few specific grindstones and seeds have been assessed on occasions (mainly in America and, more recently, Africa (e.g. Adams 2014; Harmon 2008)) but not in formats which would allow reliable transference of the data to independent external situations. Methods used in this thesis include ethnographic extraction, ethnoarchaeological review and experiment (see later in the thesis for detail). However, the data produced by this research is low-level, quantifiable and standardised and should be transposable to a wide range of research situations in Australia and elsewhere.

1.3 Background

Stone tools provide important archaeological evidence with which to reconstruct some aspects of traditional Aboriginal lifeways. Considerable research has been directed towards the technological and functional profiles of flaked stone tools, here and elsewhere (e.g. Akerman et al. 2002; Clarkson 2007; Clarkson et al. 2017; Van Peer et al. 2003). However, with some important exceptions (e.g. Adams 1999; Cane 1984; Fullagar and Field 1997; Hayes 2015; Smith 1989; 2015, 2015 et al.) less effort has been directed towards ground stone implements, especially those associated with seed use (e.g. Hayes 2015).

Perhaps this is not surprising as grindstones of all types usually form a numerically small proportion of stone artefacts obtained from archaeological excavations (Smith 1988) and grinding stones, particularly large basal millstones, are difficult or expensive to transport, examine microscopically, or store for later research (e.g. Rowan and Ebeling 2008). Even in the quite recent past, grindstones were normally allocated to some sundry classification in excavation registers and were seldom recorded in detail (Searcy 2011), perhaps not even being distinguished as top or bottom stones (Lidstrom Holmberg 1998).

Our knowledge of Aboriginal seed-grinding is much like an incomplete jigsaw puzzle providing only glimpses of the overall picture. With the aging and passing of Aboriginal informants knowledgeable about seed-grinding, rich ethnographic detail about grinding

may well remain obscure (e.g. Field and Fullagar 1998; Kraybill 1977). Fresh methods of investigation may nevertheless shed new light, such as the grinding characteristics of seeds and grindstones during controlled experiments. As Stahl observes:

as always, the archaeological record may be frustratingly incomplete as regards techniques of processing, [but] gaining an understanding of the importance of a variable often helps us to frame new questions and to look for evidence that we might otherwise have thought unimportant (Stahl 1989:185).

Important practical and theoretical research in Australia and elsewhere over recent decades has made use of Australian ethnographic seed-grinding data but the data have sometimes been incomplete or based on unproven assumptions (see later). I suggest that, as Stahl proposed, carefully formulated experiments examining the variables involved can enable key aspects of ethnographic and other seed-grinding research to be verified, supplemented or quantified. The three papers comprising the middle chapters of the thesis examine aspects of grindstones and seed-grinding from such an experimental aspect.

The first reports in English of Aboriginal Australians disparage their culture for, amongst many other supposed deficiencies, a lack of economic enterprise (Dampier [1699] in Marchant 1988). Similarly, the first European settlers also observed small-scale coastal fishing economies but had no knowledge of the complexity of the broad-spectrum economies which existed in various other parts of the country (Lourandos 1997; Mildwaters 2005). It was not until towards the middle decades of the nineteenth century that exploration moved inland from the coastal settlements and reports began to emerge of, amongst other matters, Aboriginal use of grass seed on a large scale (e.g. Mildwaters 2005; Mitchell 1848). The reports of vast tracts of arable land aroused wide-spread (and soon anthropologically destructive) interest from potential agricultural users but its economic significance for the Aboriginal owners received little notice from 19th century investigators such as Howitt (1878) or Spencer (1899) who were largely preoccupied with social issues. Some hundred years elapsed until Tindale recognised that highly developed, large scale, seed-based economies he called the Panara had existed in a great horse- shoe shaped arc through the central east, to the central north and over to much of the central west (1977). His work led to wide-spread ethnographic and archaeological interest in seeds as an economic contributor and in the ground-stone technology required for their use.

Over time, seed data from Australia, despite their limitations, were utilised in theory building at all levels in both Australia and elsewhere.

1.4 Debates in Australian Archaeology

Debates emerged in Australian archaeology about various aspects of the role of seeds and grindstones in, predominately, arid and semi-arid areas. These included, for example, the timing and pattern of the diaspora from entry in the north, adaptations to diverse environmental and ecological challenges (e.g. Smith 2013), and the timing, development and function of grinding technology (e.g. Tindale 1959; Smith 1985).

One particular unresolved debate concerns the antiquity of seed-grinding in Australia. This debate is centred on when plant processing by grinding of (mainly) seeds intensified into a crucial component of some Aboriginal economies. It is generally accepted that grindstones were an early Aboriginal technology (e.g. Clarkson et al. 2015; 2017; Hiscock 2017; Stern et al. 2013); beyond this, two opposing views dominate the debate. One holds that seed- grinding was a significant factor in some Pleistocene lifeways whilst the other sees seed- grinding as a mid-Holocene development which perhaps did not reach critical importance until as late as around 1500 years BP (e.g. Clarkson et al. 2017; Fullagar et al. 2008; 2015; Gorecki et al. 1997; Hayes 2015; Mildwaters and Clarkson 2018; Smith 1985; 1986; 1989; 2015; Veth and O’Connor 1996). A detailed understanding of grindstone morphology and the frequency of archaeological recovery of early grindstones is clearly critical to the debate.

Pleistocene grindstones recovered to date are fragmentary, small, rare and, according to a widely accepted and influential topology devised by Smith (1985), generally considered to be amorphous in design — that is, they are ‘expedient, multipurpose tools used to grind not only’ seeds but various other materials. They do not display the diagnostic formal attributes of the large millstones associated with ethnographic reports of intensive seed- use in Australia (Gorecki et al. 1997:142; Smith 1985; 1986) or (implied but not stated) elsewhere around the world.

One question to be considered is whether the very early grindstones excavated are, whilst fragmentary, actually either unexpectedly small or rare. Overall, grindstones constitute only a tiny proportion of lithic assemblages recovered from Australian excavations. In arid and semi-arid areas the proportion is only around 0.3%, with those able to be identified as seed-grinders an even lower 0.05% (Smith 1988). As these percentages include recoveries from recent times and surface artefacts abandoned when Aboriginal people left their traditional lands, it is clear that any recovered grindstone must be considered rare.

Why should grindstones be rare and why are most of those recovered either small or fragmented? Possible reasons (and some implications for other elements of current debates) include: a. From historical and ethnological records, seed-grinding was largely an activity conducted in open areas (e.g. Cane 1984; Tindale 1977). However, most excavations have been of enclosed areas such as rockshelters where the proportion of grindstones should be expected to be lower (e.g. Gorecki et al. 1997). Current strategies for excavation may not capture representative numbers of the grindstones actually used at a site (Smith 2013:13). b. Grindstones, particularly millstones, have always been considered desirable collector items by Europeans and exposed items, especially complete or ‘good examples’, were frequently removed. Elimination or reduction of important archaeological markers from sites should be factored into assessments (Gorecki and Grant 1994). c. While a muller may have occasionally been used as a pounder or a wooden spear sharpened on a millstone, as a rule a millstone and muller were operated as a tied pair. All things being equal, mullers wear out faster than their basal counterparts (e.g. Smith 1985, 1986). Unlike mullers, millstones are seldom recovered from stratified deposits (Veth 1993) but identification of a muller with diagnostic usewear presupposes the existence of its paired millstone. When a heavily worn millstone is recovered, it indicates the presence, over time, of a number of mullers; factors seldom taken into account in investigating seed-grinding from recovered site artefacts. d. In the many areas where suitable raw material was scarce, a worn-out grinding stone still represented a valuable resource. A millstone, for example, having been reduced in depth by constant pecking to resharpen, may have had the distal end

removed by flaking to allow better access to the deepening groove (Cane 1984). Only the discarded fragments may be recoverable. Alternatively, the worn millstone may have been traded onwards despite its condition (Hamilton 1980). Whether seed-grinding was an economic factor must be assessed on wider evidence than the presence or absence of more-or-less complete seed-grinding equipment. e. An exhausted millstone was sometimes used as a core to produce flaked tools when made from a suitable material (Balme 1991), or else broken — into heat- sinks to aid cooking or on the death of the owner (Smith 1986). The availability and functionality of raw material resources in an area can affect the archaeological evidence of seed-grinding present at a site. f. Large millstones had a long, sometimes generational, use-life and were curated when their owners departed from an area. By the time the camp was again visited, possibly more than a year later, natural elements may have covered small artefacts resulting from the former occupation. Large grindstones however would have been only partially buried, or were recovered from caches, and were thus reinstalled on top of other shielded items. A grindstone may have been contemporaneous with numerous spits considered separate layers by excavators. Site formation can materially affect the apparent frequency of grinding artefacts in an assemblage. g. Grinding stones were sometimes carried to and used in areas where seeds were only temporarily available and were then returned to major habitation sites when these ephemeral resources were exhausted (Veth 1993:78). This was not the usual situation, with seeds more commonly being brought into a site, however when it did occur, archaeological evidence for the practice is unlikely to be readily apparent. As Aboriginal people usually took advantage of all available ‘cheap’ (in an optimum foraging sense) resources, the presence of a desirable seed may be itself indicative of possible seed-grinding. h. Where extrusions of bedrock of suitable type were available in the landscape, flat patches on the outcrops were used in lieu of lower grindstones (e.g. Gorecki and Grant 1994; Gorecki et al. 1997; Vinnicombe 1987). Although not extensively studied, large numbers of these effectively permanent bedrock grinding patches are known. Gorecki and Grant investigated 2015 in the Croydon region of Queensland and Vinnicombe found 464 patches on the Burrup Peninsula in Western Australia. Dodson et al. (1992) analysed reports of artefacts known from sites in two states and found only 25 ‘Grinding Grooves’ recorded for Victoria (0.3%

of all artefacts) but a much greater number of 1793 (10.7%) for New South Wales. What constituted ‘grinding grooves’ is not known but may have included axe grinding grooves. Regardless, the numbers for localised areas seem very high when compared with the numbers of (functionally interchangeable) millstones usually reported. However, these figures are dwarfed by an estimate of 36000 grinding patches along Queensland’s Gregory Range alone (Gorecki et al. 1997). As well as the fragmentary recoveries from excavations, comprehensive and detailed investigations are required of the seed-grinding implements that are fixtures in the landscape. i. Seed-grinding is a relatively simple technology, either inherited from anatomically modern humans (AMH) in Africa, possibly as far back as around 250 kya (thousand years ago) (Van Peer et al. 2003) or periodically reinvented over the period of the diaspora. It was a known technology in the early occupation of Australia some 65 kya (Clarkson et al. 2017) despite the original arrivals likely being adapted to a marine or coastal economy with limited reliance on grinding (e.g. Bowdler 1977; O’Connor and Chappell 2003). Using this known seed technology as an insurance against economic failure may have facilitated initial movements away from the coast. However, seed use may not necessarily have continued once familiarity with new environments was established and higher ranked resources were assimilated into diets. Comparable situations may have occurred in other circumstances such as fissioning of a group forcing reliance on lower ranked resources in the reduced new territory (McNiven 1995). Accordingly, while episodic grinding only may be reflected in the archaeological record, its possible critical importance should not be overlooked. j. Climate has fluctuated markedly in the period since first occupation (e.g. Dodson et al. 1992; Hiscock and Kershaw 1992) and with it, grasslands have expanded and contracted. For example, in Western Australia, Semeniuk (1995:253) has suggested that the arid band presently centred on the Pilbara has, over the past 7000 years, migrated north to this position at a rate of around 100 kilometres each thousand years. Climatic conditions at the period under investigation need to be known as evidence for seed-grinding can only be expected where and when environmental conditions allowed or necessitated seed use.

k. Finally, grindstones are subject to the problem of assemblage size and sampling as discussed in a series of papers by Hiscock (e.g. 2001). Using Hiscock’s theories and the sites surveyed by Smith (1988), Gorecki concluded that:

If most of the grindstones, including the subset of seedgrinders, come from more recent levels and these are only a minor proportion of all retouched artefacts, then the probability of recovering seedgrinders in earlier levels where assemblages are much smaller would be very low indeed (Gorecki et al. 1997:146).

Taking all these factors into consideration, it should not be surprising that recovered grindstone artefacts are often only fragments and even the fragments will likely be small in number. Unless abandoned, few complete grindstones are likely to have entered the archaeological record.

If the scant evidence available is to be correctly interpreted, it is important that fine-grained information on grindstones — morphology, use, wear, productivity, etc. — is available.

The second arm of the problem, the seeds which were ground, requires equally detailed data to enable such matters as resource selection, ranking in productivity models and female labour investment to be determined.

1.5 Global Directions in Grindstone Research

The volume of groundstone or, as proposed by Adams (2002) macrolithic, research is small compared with the effort that has been devoted towards flaked stone investigations (e.g. Rowan and Ebeling 2008). However, as discussed in detail later, a world-wide interest in grinding as it contributes to ‘baseline technological traditions’ has progressively developed over recent decades. (For the purposes of the thesis, non-seed processing implements are not addressed in detail). Much of this research has been directed towards establishing how grinding implements were used and what materials were processed. This has largely been accomplished using microscopic and chemical analysis of traces of usewear and residues remaining on the artefacts compared with known experimentally derived reference materials. Researchers in Australia have been significant contributors to the development and acceptance of usewear and residue analysis. (The number of papers

dealing with this specific area precludes a full listing but the following will provide an overview — Adams 2014; Delgado-Raack and Risch 2009; Dubreuil 2001; Fullagar 1991; Fullagar et al. 2012; Hayes 2015; Loy 1992; Piperno et al. 2004; Smith et al. 2015).

While usewear and residue analysis research is increasingly successful in determining the biomechanics of use and identifying the plants ground to taxa, and sometimes species, level, more detailed knowledge is needed of seed-grinding economies. Limited research on other aspects of the grinding puzzle is being undertaken around the world. Table 1 shows a sample of papers published in English — please note that the papers included in the table are often only representative of the body of relevant work published by that researcher. ‘Area’ is the most specific geographic sector identified.

Researcher Publication Area Relevant Key Phrases Comments Date Content Adams, Jenny 1993 - 2014 Southwest Formation and Technological evolution; Causes A pre-eminent authority on USA use wear and of wear; Tribology; Research groundstone (macrolithic) artefacts) motor actions. design; Terminology. and has published widely. 2014 book is a comprehensive technological guide to ground stone research. Bartlett, Katherine 1933 Southwest Metate Grindstone morphology; Seminal work on seed-grinding using USA efficiency. Formation use wear; Motor metates and manos discussing motor actions of grinders; Grinding actions of grinders and wear formation efficiency of metate and mano and relative efficiency of the types. grindstones. . Brokensha, Peter 1975 Central Times to Seed processing observations; Often quoted research which gives Australia process seeds. Preparation of traditional damper times for various tasks involved in seed from grass seed; Time required processing. for procedures. Cane, Scott 1984 Central Grindstone Formation wear and morphology Important ethnoarchaeological Australia morphology of seed-grinders; Seed research on the processing of various and seed processing from collection to seeds by native millers, including time processing. grinding; Times required to grind to grind, and formation wear of various seeds. grindstones. David, Nicholas 1998 Nigeria Development Bedrock grinding hollows; Contrasts the historical and of grindstone Grindstone morphology; technological development of grinding formation Formation use wear; Quern hollows with present grain processing usewear comparative functions; Mortar using ethnographic manufactured uses; Biomechanics. querns and mortars. Delgado Raack, S. & 2009 Africa; Technological Morphological variability; Summary of global cereal grinding Risch, R. Global analysis of Grindstones as indicators of including the possible benefits of using seed-grinding subsistence production; Multi- a wooden muller on a narrow, convex tools. functionality of grindstones; stone grinding slab to separate cereal Upper and lower grindstone bran from flour. compatibility;

Eitam, D.; Kisley, M.; 2015 The Wild barley Pre-agriculture; Late Natufian Experimental demonstration of Late Karty, A.; Bar-Yosef, O. Levant processing; culture; 12500 years ago; Natufian capability of producing groats Mortar Sedentary hunter-gatherers; and fine flour from wild cereals. functionality. Conical mortar. Haaland, Randi 1995 Africa Grindstone Grindstone manufacture; Investigation of wild sorghum manufacture Cultivation of wild sorghum; Plant cultivation several thousand years and wild seed domestication. before morphologically visible grinding. domesticated forms emerged. Hamon, Caroline 2008 Europe Grindstone Replication of Early Neolithic Early Neolithic sandstone tools from usewear and grindstones; Sandstone as raw north-western Europe duplicated and function. material; Functional identification; used to experimentally process a range Usewear Reference Collection; of materials. Use wear on both sets compared to suggest original function of the archaeological examples. Hard, R.; Mauldin, R; 1996 Southwest Mano size. Mano size: Stable carbon isotope Increased mano size is one of the lines Raymond, G. USA ratio: Macrobotanical remains. of evidence of maize dependence Hayden, B. 1987 Guatemala Metate Reasons for choosing a particular Frequently cited ethnoarchaeological manufacture. stone type: Positive qualities of paper on contemporary metate and vesicular basalt: Effects of mano manufacture. repecking on use-life. Horsfall, G. 1987 Western Design theory Never a "best" solution only a Artifacts, including grindstones, made USA in "satisfactory" solution: to solve an activity-related or adaptive archaeology. Socioeconomic constraints lead problem. to grindstone variation: Tripodal metates confer social prestige. Huffman, Thomas 2006 South Seed type and Sorghum and maize processing; Different quern / handstone sets are Africa grindstone Grindstone morphology. parts of separate technologies evolved specialisation. Specialisation of maize quern. to accommodate the differing requirements of sorghum and maize seeds. Lidstrom Holmberg, 1998 Sweden Grindstone Technology of Mesolithic- Formation of upper and lower Cecilia morphology Neolithic transition; Saddle- grindstone specialised morphology and function shaped querns and loaf investigated by experimental grinding. . handstone; Standardisation; Use wear.

Mauldin, R. 1993 New Ratio of Mano area the best predictor of Mano and metate grinding areas Mexico grinding area grinding time: Calculation of positively correlated with time required USA to grinding Correlation Coefficients supports: to grind. time. As grinding area increases, time to complete shrinks. Nixon- Darcus,L.; 2017 Ethiopia Manufacture Manufacture of grindstones; Comprehensive ethnoarchaeological D'Andrea, A. and function of Comparison of quern types; Use assessment of modern grinding in querns and wear and maintenance. Surface Ethiopia including manufacture, use handstones. roughness (grit) matched to and maintenance of querns and seed. handstones. O'Connell, J.; Hawkes, K. 1981 Central Time to grind. Seeds in Optimal Foraging Seminal work situating Australian Australia model; Demonstration of grinding native seeds in Optimal Foraging mulga; Extrapolation and model including estimates of times estimates of grinding times. required to grind seeds. O'Connell, J.; Peter, K; 1983 Central Time to grind. Seed preparation including Adjustment of estimated time to grind Barnard, P. Australia collection, parching and pre- based on a brief grinding experiment. cracking; Descriptions of seeds; Experimental grinding. Pastron, Allen 1974 Mexico Manufacture Metate and mano types and Ethnoarchaeological observations of and function of morphology; Manufacture; Male the Tarahumara of Mexico and metates & and female labour contributions; manufacture and use of metates and manos. Use-life. manos. Robitaille, Jerome 2016 Ethiopia Handstone Granite grindstones manufacture Handstones and querns serve the function and by men and women; Absence of function of the not-known mortar and versatility. mortar and pestle; Macroscopic pestle and can function as both formation wear; use-life. specialised and multi-functional implements. Schneider, Joan 1996 Arizona Groundstone Processing of plants; Quarrying Analysis of quarrying and specialised USA implement of raw materials; Manufacture of production of lower grindstones and manufacture implements. pestles. Searcy, Michael 2011 Guatemala Manufacture to Basalt as raw material; Quarrying Comprehensive ethnoarchaeological discard of and manufacture of implements; investigation tracing the use-life of manos and Gender- made by men and used manos and metates from quarrying of metates. by women; Extended raw material to eventual discard. generational use-life.

Shoemaker, A.; Davies, 2017 Kenya Cereal Raw material quarrying, Comprehensive ethnoarchaeological M.; Moore, H. processing manufacture and maintenance; investigation tracing the use-life of with stone and Formation usewear; Surface grit grindstones from quarrying of raw wood important for sorghum and maize material, manufacture, maintenance implements processing; Wood mortar and and eventual discard. stone pestle. Smith, M. A. 1985 & Central Classification, Classification of grindstones; Influential ethnoarchaeological papers 1986 Australia morphology Seed-grinding toolkit; Morphology suggesting classification of grindstones and function of and formation wear of - seed-grinders, formal, specialised and grindstones. grindstones. Processing of amorphous types- and describing their seeds. use in the processing of seeds. Importance of functional surface. Smith, M.; Hayes, E.; 2015 Australia Residue Exhausted seed-grinder; Focus of paper is on analysis of Stephenson, B. analysis and Formation usewear; Microscopic residues from an exhausted experimental use-polish and usewear; Residue ethnographic millstone but also reports grinding. and starch analysis; Time to a brief seed-grinding experiment. grind. Stone, Tammy 1994 Arizona Importance of Influences of availability of raw Significant decisions about grinding USA local and materials; Reasons for determined by raw material available. imported raw preference for vesicular basalt; Vesicular basalt important for materials. Macroscopic formation usewear manufacture of metates and imported and maintenance; at high cost but raw material less critical for manos. Wilcox, G.; Stordeur, D. 2012 Syria Flotation Wild barley and wild rye; Evidence for large-scale wild cereal analysis of Cultivation of non-domesticated processing and storage around ancient seeds seeds. c9000ya, 1000 years before domestication. Wright, Mona 1993 USA Experimental Replicated manos and metate; Experiment recording production of production of Experimental maize grinding; maize flour and weight loss of metate maize flour. Maintenance of surface and manos from seed-grinding and roughness (grit) by pecking. sharpening of implements Table 1. A sample of global grinding research (excluding papers based predominately on use-wear and residue analysis).

Examination of Table 1 confirms that grinding research is now a complex global endeavour. For example, subjects such as raw material qualities and acquisition methods and the manufacture, morphology, function, use and maintenance of grindstones are all aspects receiving attention (see the table for references). Often investigations are supported by experiment. Plant processing is also receiving attention but, as mentioned elsewhere, few studies record actual grindstone efficiency or seed productivity. The seeds investigated are largely the widely used domesticated cereals or their wild forbearers. Surprisingly, only one paper identifying seeds by flotation was located (Wilcox and Stordeur 2012).

A number of issues raised are unusual and intriguing. An assumption in African archaeology that ‘the proliferation of grinding-stone tools’ can be ‘used as proxy evidence for … the development of food production’ is challenged as seed processing is shown to have a time depth extending back before domestication of the seeds (Shoemaker et al. 2017:416). Plant processing in Africa using wooden implements in conjunction with, or in substitution for, traditional stone raw materials is discussed by Delgado Raack and Risch (2009) and Shoemaker et al. (2017). Formerly, discussion of seed processing with wooden implements was largely restricted to ethnographic North America (e.g. Kroeber 1925). Delgado Raack and Risch record a specialised grinding pair where a wooden muller used with a long, thin convex stone millstone enabled flour and bran to be separated out on the sides of the millstone (2009).

Although attracting little attention, in Australia, there are a number of references to plant processing (including that of seeds) being performed using wooden implements. For example, Bulmer reported that nardoo was ‘pounded up in a wooden trough’ (1887:15), Cleland and Johnston noted grass seeds being ‘pounded dry in’ a wooden dish, ‘with a stone’ (1937:208), and Gorecki et al. state that in ‘areas where stone is near impossible to obtain … then other platforms like wooden dishes become the local versions of grindstones’ (1997:142). Because of their impermanence, wood artefacts seldom feature in archaeological assemblages but

their possible contribution to seed-grinding economies in Australia warrants detailed investigation.

1.6 Methods

While methods included analysis of historical, ethnographic and excavation reports, primarily an experimental methodology is adopted. The data relating to grinding implements and seeds were obtained by carefully controlled experiments which closely followed the recommendations of various authorities on experimental archaeology as detailed later.

1.7 Seed-Grinding in Theoretical Context

In the mid-nineteenth century, the great geological time-depth of the earth was gaining acceptance and questions were being asked about certain objects imbedded in layers of obviously ancient soils. Were these objects naturally produced ecofacts or could they have possibly been the product of human creativity? Successful experiments were able to reproduce these artefacts and propose use-functions which helped convince scientists and others that humans were an evolved species and had a history that was much older than the theological dogma of the time indicated (Coles 1979). As such, long before the emergence of archaeology as a discipline, the perhaps earliest ‘archaeology’ was both ‘Darwinian’ and ‘experimental’ in nature. An evolutionary theory and an experimental methodology were predecessors of later orientations, regardless of how evolutionary concepts and experimental results may now be interpreted from within particular theoretical viewpoints (Outram 2008).

The distinctions between the various ‘Darwinian Archaeologies’ are sometimes finely drawn. However, with its evolutionary and ecological parentage, scientific orientation (including a theory of mind) and emphasis on cultures and behaviours, Human Behavioural Ecology (HBE) appeared to provide a framework which could both guide the experiments and situate the results within a larger research tradition (Bentley et al. 2008:154-7). In short, I understand HBE to hold that ‘diversity in human culture

and behaviour’ (forms of phenotypic plasticity) ‘results from selection which has shaped our ability to adapt to diverse social and ecological conditions’ by way of an evolved set of decision rules - rules which constitute the adaptation, rather than do the behaviours themselves (Bentley et al. 2008:154).

A brief review of ecological approaches in Australian archaeology suggests that many of the key papers discussing issues of importance to this thesis, although not always explicitly stated as such, are generally ecological in perspective (Pardoe 1994; Veth et.al.2000). The ‘link between evolutionary models of human behaviour and social organisation is through ecology’ (Pardoe 1994:189). A human behavioural ecological (HBE) orientation was thus considered an appropriate theoretical framework in which to situate the experimental study and highlight the critical decisions which people had to make when ground seeds were to form a substantial component of a traditional economy.

The research traditions in which HBE concepts are often invoked include areas as, for example only, the ‘broad spectrum revolution’, rational choice models and various foraging theories. More recently, Cultural Niche Construction (CNC) has been suggested as a theory capable of helping further explicate such evolutionary issues (e.g. Florin and Carah 2018; Rowley-Conway and Layton 2011). Regardless, the data requirements of such evolutionary models ‘often are extensive and difficult to meet’ (Winterhalder and Smith 2017:17). Where seed use is involved, such investigations could arguably make good use of additional transportable, quantitative data. The aspect of availability of data, or sometimes the lack of it, is a central issue of the thesis which could not be fully addressed in detail in the individual papers. It is briefly discussed below.

Over the past four or so decades, HBE has become an important investigative arm of archaeology and has attempted to address a number of archaeologically important issues using formal predictive models developed from areas such as optimisation analysis and game theory (e.g. Clarkson 2007). Matters addressed have been as diverse as, for example, functional or adaptive patterns in optimal foraging, resource transport, changes in technology, the origins of agriculture and various material correlates of human social organisation (Bird and O’Connell 2006:144). An influential

prediction of HBE was that, regardless of conventional explanations for the ‘broad spectrum revolution’, which was generally adopted after the terminal Pleistocene (but later in Australia) (Edwards and O’Connell 1995:769), the new resources added to a diet, including ground seeds, usually ‘had very high handling costs relative to energetic yield; that is, they were low ranked compared with previously favoured resources’ (Bird and O’Connell 2006: 148). The seminal work of O’Connell and Hawkes (1981) and O’Connell et al (1983) with the Alyawarra people of central Australia became a corner-stone supporting a number of the models. The widespread adoption of seed-grinding generated debate concerning the economies of both pre and post ‘broad spectrum’ eras. In addressing technological aspects of these questions, developers of formal treatments found difficulties in quantifying variables, ‘especially those related to toolstone utility and the costs and benefits associated with the use of different implement forms’ (Bird and O’Connell 2006: 152; Simms et al. 2013). Whilst a relatively large corpus of information was available for flaked stone tools, information concerning the ground stone implements of seed- reliant economies was comparatively scarce and broad grained. Surprisingly often, researchers made use of a limited selection of Australian ethnographic source-side data such as that of O’Connell and Hawkes (1981 above), Brokensha (1975) and Cane (1984) despite the fact that: many organisms (certainly humans) have the evolved capacity for rapid adjustments favouring advantageous behavior, [and] they are likely to exhibit predictable short-term, often ‘real-time’ adaptive responses to many social and ecological features of their environment (Bird and O’Connell 2006:145). This combination of rapid adaptive technological responses and only broad-grained information about that technology, suggests that the predictions developed could possibly contain significant degrees of uncertainty.

For example, Bright et al. (2002), in their ‘Tech Investment Model’ from the Great Basin area of America, proposed that, as the availability of high-ranked animal resources declined, diet breadth increased. Importantly, this required development effort to be diverted from flaked stone hunting technology to improvement of the grinding stone technology required to process low-return seeds. The general model is influenced not only by the American data of Simms (1985) but also by the Australian data of O’Connell and Hawkes (1981). Whilst Simms provided detailed

estimates of hunting costs, the seed component of his research did not include the grinding of the seeds collected. Similarly, the seed-grinding times of O’Connell and Hawkes (1981) contain a large element of estimation. As Madsen and Simms point out, ‘the success or failure of theoretical models in adding new insights hinges on the ability to recognize past behavior in the archaeological record’ (1998:280) and past behaviour can only be inferred with any degree of confidence if appropriate data are available (Adams 2014:4-5). Further development of such a model would require detailed, fine-grained knowledge of the technological options available and the returns from a range of fully processed seeds.

As more fully discussed in the next chapter, there may be further, and other, methodological problems involved in applying ethnographic data sets beyond their initial ambit. Comprehensive, quantifiable and transportable fine-grained experimental data can support and supplement limited information from ethnographic sources (Clarkson and Shipton 2015).

1.8 Experimental Archaeology

Not all archaeological experiments can be considered experimental archaeology (Schiffer et al. 1994). For example, it may merely be that ‘a copy is made of an original artifact with attention paid only to its visual appearance for display purposes’ perhaps as a setting in a museum (Coles 1979:36) or as part of some re-enactment of past life (Outram 2008:3). Rather, modern experimental archaeology involves a rigorous application of scientific research methods which include ‘basic procedural rules that are applicable to all [scientific] experiments’ (Coles 1973:15), a hypothesis that can be tested to see if it can be falsified (Clarkson and Shipton 2015; Outram 2008) and should ‘provide or enhance analogies for archaeological interpretation’ (Mathieu 2002:1) . This ‘objective’ orientation itself has been criticised by post- modern theorists (e.g. Shanks and Tilley 1987). However, if experimental archaeology is to make an ongoing contribution to the discipline, it is difficult to see how it could otherwise be meaningfully undertaken. Numerous ‘rules’ have been advocated for conducting experimental archaeology (e.g. Adams 2010; Clarkson and

Shipton 2015; Coles 1979; Mathieu and Meyer 2002; Outram 2008; Schiffer et al. 1994).

This research project has been undertaken with such guidance prominently in mind. Some of the proposed rules and their applicability to this research are discussed below- 1. Clearly define the experiments and frame questions in answerable formats (Adams 2010:143; Mathieu and Meyer 2002; Outram 2008:4). The aims of the experiments were clearly stated in an easily answered form, for example: What are the outputs of ground product for grindstones of varying morphology? 2. Decide on a strategy — confirm or confound existing research or explore new aspects (Adams 2010:144). While various seed grinding experiments have been conducted around the world and provide the base strategy for the experiments, none had the precise problem orientation required. However, some student seed-grinding has been conducted at the University of Queensland (Clarkson and Shipton 2015). It was decided to formalise and extend these preliminary experiments. 3. Provide full details of materials used and methods adopted (Outram 2008:4). Materials and methods were described in detail. 4. Select Factors and Response Variables and ensure the parameters are appropriate (Adams 2010:144; Outram 2008:4). The variables examined were appropriate and largely self-evident. 5. Use materials which would have been originally available or, if compromises are necessary, provide full justification for the materials used (Coles 1979:46; Mathieu and Meyer 2002; Outram 2008:4). Traditionally utilised medium grade sandstone was used to reproduce the grindstones. A modern artificial industrial compound abrasive, for example, would not have provided equivalence. Native seeds were mostly used. A few domestic seeds were used as proxies and these were explained and evaluated in detail. 6. The methods and tools used should not exceed presumed traditional competence (Coles 1979:46). An electric grinder was used to manufacture the replicated grindstones but it was the functionality of a particular grindstone

morphology or seed type that was being investigated, not the time required to manufacture the grindstones themselves, so this precept was not violated. 7. A system to facilitate accurate and detailed quantification, recording and analysis should be established (Adams 2010:145); Coles 1979:46). Word- processing, spreadsheet and statistical computer programmes, Vernier callipers and electronic timers and scales were used. Physical dimensions of the grindstones and functional grinding areas including initial wear depths were fully logged as were physical dimensions of the seeds. Times and weights were measured and recorded. Comments on unusual individual grinding episodes were noted. 8. The models produced must be accurate representations and the effects of any scaling must be assessed (Clarkson and Shipton 2015; Coles 1979:46). All grindstones were manufactured as full size models confirmed as accurate from archaeological examples and ethnographic images and descriptions. 9. The experiment must be performed, recorded and the elements of the experiment repeated sufficient times for anomalies to be recognised. Repetition of specific actions is important and series experiments provide greater understanding (Adams 2010:145; Clarkson and Shipton 2015; Coles 1979:46; Mathieu and Meyer 2002). Individual grindings were recorded on a spreadsheet and repeated a number of times. Grinding series (e.g. by grindstone or seed type) were compiled and subjected to statistical analysis. 10. During the experiment, ongoing problems must be examined and improvements or improvisations implemented (Coles 1979:47). Where problems were encountered individual grindings were repeated to confirm the anomalous result or the grinding was eliminated if the result proved non- repeatable. Further seeds were added as the importance of comparative series became apparent. Other improvisations were not necessary and improvements will be implemented by way of future extensions of the experiments (Mathieu and Meyer 2002). 11. Experimental results must not be taken as proofs (Coles 1979:47). ‘Equifinality is an unavoidable problem in experimental archaeology’ (Clarkson and Shipton 2015:160) as, for example, functional success as anticipated does not rule out alternative uses or the success achieved may have been also possible from quite different approaches. Acceptance of this caveat has

been made clear in the text. Where possible, alternatives were explored. For example, assessments of the crushing ability of a millstone and muller set, as opposed to a mortar and pestle, were made as was the capability of a mortar and pestle to produce fine flour. 12. A detailed and honest evaluation of the experiment, including subjective personal aspects must be made (Adams 2010:146-7; Coles 1979:47). As far as is possible, an objective evaluation was made and both positive and negative results, plus desirable future enhancements, were evaluated and recorded. 13. Communicate results. A report on early exploratory seed-grinding was presented to the Australian Archaeological Association Conference at Coffs Harbour in December 2013. Chapter 2 has been published in The Artefact in 2018 and Chapters 3 and 4 in the Journal of Archaeological Science: Reports in 2018 and 2020 respectively.

A further, and critical, point has been made about the differences between archaeological experiments and experimental archaeology by Schiffer et al. (1994). ‘Archaeological experiments tend to be one-shot affairs, done in isolation and concluded before there has been an adequate opportunity to conduct the study rigorously. Such experiments seem like pilot studies’ (1994:198). They do not populate a historical or environmental framework and lack ‘academic context’ (Outram 2008:1980). In contrast:

a program-based experimental archaeology entails the creation of a new technology and the establishment of a technological tradition. In experimental archaeology, individual experiments do not exist in isolation, but draw expertise and technology from the program’s tradition and, in turn, contribute to its elaboration. More importantly, the findings of one experiment are nested within families of related principles (correlates) that, together, furnish a foundation for explaining technological variation and change (Schiffer et al. 1994:198)

Seed-grinding technological traditions have been established in a number of parts of the world over recent decades. These experiments have been designed to fit within and extend these traditions, both in Australia and elsewhere.

1.9 Introductory Conclusion

A number of calls have been made for further experimental grinding data to be added to the corpus of seed-grinding knowledge (e.g. Adams 2002; Fullagar 1994; Simms 1985; Veth et al 1997). However, experiments are perhaps most valuable as a supplement to real-world experience.

In this respect, grinding knowledge is becoming difficult to obtain in ethnographic settings. Aboriginal people with everyday experience of seed-grinding are, due to the aging of these prospective informants, becoming rare. Further, when seeds are now ground, it is often as a social or ceremonial activity which can affect which seeds are collected; there is no certainty that the seeds traditionally depended upon are those used. It is critical that strenuous efforts are made, here and elsewhere, to record what precious knowledge is still accessible. Regardless it is clear that, in the future, more reliance will, of necessity, need to be placed on highly detailed experimental studies. Such experiments will need to clearly define the technological parameters of the implements utilised, the types, treatments and characteristics of the seeds ground and provide quantified transportable outcomes capable of being integrated into a variety of scenarios.

The experiments discussed in subsequent chapters meet these requirements.

1.10 The Papers

The first paper (Chapter 2) reviews current knowledge of seed-grinding, summarises cases used in theory building and examines possible methodological problems with some of the data utilised.

The second paper (Chapter 3) examines the productive efficiency of various grindstones by way of experiments with a number of replicated grindstones and three proxy commercial seeds.

The third paper (Chapter 4) investigates the grinding profiles of a number of native seeds and the domesticated seeds processed in Chapter 3, using a single replicated millstone and muller pair.

1.11 References

References for the Introduction and Conclusion appear after the Conclusion. References for Chapters 2 to 4 appear within the publications.

2 Chapter 2 - Seed Grinding in Traditional Aboriginal Australia

2.1 Introduction

The paper largely provides a concise background to Aboriginal seed grinding: a wide-ranging, complex and contentious subject. It examines existing research and assesses possible effects of omissions, imprecision and assumptions on outcomes.

2.2 Methodology

Methodology was by way of review and critique of existing published ethnological, ethnoarchaeological, archaeological and experimental papers.

2.3 Conclusion in summary

Fundamental problems with seed grinding are not well understood and, as a consequence, existing practical and theoretical assessments depending on grindstone efficiency are only able to address problems in broad terms. Greater clarity should be possible if the underlying technological factors affecting seed- grinding are known.

The relevance and value of ethnoarchaeological studies of grinding are acknowledged and future research is urgently encouraged while it is still possible. The fact that few of the grinding experiments reported were performed by women who had actually ‘ground for a living’ is raised as a possible caveat.

The paper provides an appraisal and summary of a number of complex themes in seed grinding research. It is hoped that it will prove to be a viable reference point and substitute for otherwise lengthy discussions of such issues.

2.4 The Paper

The paper was published as: Mildwaters, John. 2018. Seed-grinding stones: a review from a mainly Australian perspective. The Artefact 2016. Volume 39, pp. 30- 41. The paper, prior to final typesetting, is as follows:

Keywords Australia – Grindstones – Morphology – Productivity – Seeds.

Title- Seed-grinding stones: a review from a mainly Australian perspective. Author- John Mildwaters Photographs- John Mildwaters

Abstract The role played by grindstones in converting seeds to ingestible products in Australian Aboriginal economies is presently only understood in broad, general terms. Important theoretical work has been based on reported grinding performance but the data have not always been thoroughly interrogated for accuracy or aptness under the specific circumstances being considered. Using ethnographic and excavation reports, this paper assesses the ethnographic literature, reviews the baseline factors that affect grinding performance and suggests suitable areas for further exploration by experimentation with replicated grindstones.

Introduction Seed processing activities have provided important avenues of investigation into a number of areas of concern to archaeologists. Theory building, in and outside of Australia, addressing foraging, diet breadth and staples, nutrition and digestibility, land occupation, intensification, division of labour, and technological time-depth, organization, embeddedness and investment has utilised, directly or indirectly, Australian ethnographic seed-grinding information (e.g. Binford 1979; Bird and Bliege Bird 2005; Bright et al. 2002; Dubreuil 2004; O’Connell and Hawkes 1981; Simms 1984, 1985; Stahl 1989). In particular, the role of grindstones was identified as being of importance, and particularly so, in many of the drier areas of the world, including Australia with its extensive grasslands and acacia shrublands (e.g. Bartlett 1933; O’Connell 1977; Simms 1984; Smith 1985, 2013; Tindale 1977).

What exactly is meant by ‘grindstone’ is a matter of considerable confusion in the literature, and beyond detailed discussion here: a complete and updated typology is urgently needed for Australian grindstones. For the purposes of this paper, ‘grindstone’ is used as a general term for the stationary bottom, or basal, stone of a grinding pair and, whilst the term would normally also include the active top-stones (and other grinding implements), these are referred to more particularly as mullers or topstones or, when associated with a mortar, as a pestle. A grindstone may not necessarily equate to a more specialised ‘seed-grinding’ stone, or millstone, utilised to reduce seeds to meal. Examples of ethnographic and replicated millstones and mullers are shown in Figure 1. More specific terms are defined in context throughout this paper.

26 A. B. Figure 1. A . Anonymous traditional heavily worn millstone and muller (author’s collection). B. Replicated millstone and water-worn pebble muller.

Whilst Australian seed-grinding stones had elements of multi-functionality, ethnographic observation suggests that they were predominantly the primary extractive implement used by women to produce food products ready for consumption or cooking, and not usually to achieve some intermediate purpose such as the construction of another implement (Cane 1984:137). As such, they are of special importance as they sometimes provide direct evidence in the archaeological record of their role in past lifeways, for example, by retaining food residues and recording use-wear (e.g. Fullagar et al. 2008; Hayes 2015). If, as has been suggested, female hunter-gatherers are largely invisible archaeologically, and if they were primarily responsible for the preparation of ground food products, then grindstones provide a possible means to examine a significant portion of their labour in seed-grinding economies (Bird and Bliege Bird (2005:81).

Clearly, the ability of a grindstone to produce food is of critical importance but examination of the Australian and overseas reports specifically addressing grindstone productivity reveals some serious methodological problems. Many of the assessments have been based on limited observations or very brief tests and a number appear to repeat previous work. In addition, many do not identify the morphological parameters of the seed-grinding stones utilised although a few notable exceptions exist (e.g. Bartlett 1933; Cane 1984; Haury 1950; Hayes et al. 1981; Johnson 1963, 1964; Veth and O’Connor 1996). By omission, an implication of sameness of performance of seed-grinding stones has developed - that is, researchers have not considered it necessary to detail the morphology of seed- grinders - despite the physical differences between various implements (e.g. Allen 1998; Hard et al. 1996).

Grindstone analysis potentially addresses basic research questions ‘about the specific attributes that allow archaeologists to recognize tool manufacture, use, maintenance, and discard’ (Adams 2014:129). However, even the fundamentals of food grinding - the baseline technological factors (Adams 2002:17), are still only broadly understood. Use wear and residue analysis are receiving attention in the literature (e.g. Adams 1999, 2014; Fullagar 2006, et al. 2015; Hayes 2015) but other aspects of function, including grinding efficiency, need to be explored in detail (Adams 2014; Simms 1984:44, 50). As well as productivity, these fundamentals influence numerous other factors in a seed-grinding industry including labour inputs, raw material acquisition, investment in technology, maintenance and discard regimes and, importantly, the traces they leave in the archaeological record. Technological

27 dynamics ‘feed back into questions of higher theory’ (Beck et al.1989). Unless these baseline factors are carefully established, a real risk exists of accepting ‘dubious assumptions about how performance characteristics are affected by technical choices’ (Schiffer and Skibo 1987:615). A wide-ranging and ongoing debate between Smith (1985, 1986, 2004; 2015b), Veth and O’Connor (1996), Gorecki et al. (1997) and others concerning various aspects of grindstone technology including the status of formal and amorphous grindstones, and ‘seedgrinders’, has been a point of contention for some time in Australian archaeology. Part of the debate concerns what implements were used in seed- grinding. In essence, Smith suggested that only four formal artefacts - millstones (the larger, passive basal stone sometimes assumed to be a ‘specialised’ seed-grinder), mullers (the active topstone), and mortars and pestles - were regularly used whereas others proposed ‘that there is no valid functional distinction’ between Smith’s formal and amorphous types and various other implements (Gorecki et al. 1997:144). These might include bedrock grinding patches, pitted anvils and possibly axe-grinding grooves (Pardoe 2015). If Smith’s restrictive definition is able to be expanded, a better understanding of all the components involved could result in seed-grinding being recognised as being ‘more abundantly represented in the archaeological record’ than is presently appreciated (Gorecki et al. 1997:144) and possibly allow variations within seed-based economies, including the component of female labour, to be identified (Smith 2004:183). Smith himself has made a move in this direction by establishing that general purpose tjiwa grindstones from the Western Desert were also used for processing seeds (Smith 2015b:117).

Another of the contentious issues in Australian archaeology is the apparent mismatch between the early use of seed foods (c. 65000 years BP) (Clarkson et al. 2017) and the much later adoption of the large specialised millstone sets in the mid to late Holocene (Smith 1985; 1986). The archaeological record (for example, see Smith 1985, Hayes 2015) suggests that Pleistocene and Early Holocene grindstones were probably quite small and evidence for mortars before the Late Holocene is ambiguous (Smith 1985; 2004; 2013:332.). It is unclear whether this early, small grindstone technology would have enabled significant use of grass seeds.

The reported high labour costs of the grinding component of a seed-based economy have led researchers to suggest that seeds would have been low-ranked products in a foraging economy (e.g. Cane 1987; Edwards and O’Connell 1995; Hawkes and O’Connell 1981; O’Connell and Hawkes 1981). However, an alternative view suggests it is possible that the actual grinding of soft seeds ‘represents a small fraction of total processing time’ (Simms 1985:119). Important elements of foraging theory have been developed using ethnographic examples of seed-grinding labour costs including O’Connell and Hawkes’ influential study of the Alyawarra seed use as discussed later (1981). It is essential that the situation be clarified.

This paper assesses the available grinding literature, identifies a number of the baseline factors that pose problems or require investigation, and identifies areas where it is believed worthwhile knowledge could be gained by replication studies.

28 Seed-grinding Stones through Time The temporal record of seed-grinding and its associated technology in Australia is not clear. It has been hypothesised that seed-grinding would have been a colonising technology in Sahul (the Pleistocene conjoined landmass of Australia, Tasmania and New Guinea). Clarkson et al. have presented evidence for seed-grinding at 65,000 years BP (Clarkson et al. 2017; Fullagar et al. 2015; Hayes 2015). However, the picture is still far from agreed. Most recovered implements that can be securely established as seed processors date only from the mid-Holocene (with a possibly important intensification of use as late as 1,500 years BP) through to ethnographic times. (Smith 1986; 1989). A small number of grindstones, mainly topstones, have been recovered from Pleistocene age sites such as the Willandra lakes in western New South Wales and Arnhem Land in the Northern Territory (Balme 1991; Clarkson et al. 2015). It should be noted that other sites such as Serpent’s Glen in the Little Sandy Desert of Western Australia, show no evidence of early seed- grinding despite large quantities of grinding implements from later periods being present (Veth and O’Connor 1996; O’Connor et al. 1998). Under the influential typology proposed by Smith (1985), these are generally considered to be amorphous, multi-purpose grinding tools rather than specialised implements for high- volume processing of seeds. They perhaps represent evidence for seed-grinding but not necessarily for a seed-grinding economy (Smith 1985; 1986; 2015a:176). Tindale suggested that there was actually an evolutionary sequence of grindstone development from pounders and anvils in the Pleistocene to the large grooved mill sets of ethnographic times, although he did not fully develop his proposed sequence (1959).

Regardless, as Allen points out, the most significant variation between Pleistocene and Holocene technology may be in how people organised their societies to take advantage of the changes (1998). The morphological differences may not be extensive but do need to be established (Smith 1989:312). Smith suggested that seed-grinding may not have involved the introduction of new implements but rather a social reorganisation of existing technology for ‘the adoption of a new mode of subsistence’ (Smith 1993:43) - the specialised exploitation of grass seeds.

The establishment of a progression of differential development in grinding technology elsewhere in the world has proved informative (see, for example, Adams (1993,1999) and deBeaune, -who proposed an evolutionary sequence for pounding and grinding stones based on the cognitive demands of varying types of plant processing (2004). In some parts of the world, formal grindstones came to dominate assemblages; that is, morphology became relatively standardised depending on use (e.g. Adams 1993). At least an element of intentional design beyond mere decoration or weight saving was apparent in what were semi-manufactured products rather than simply the by- products of varying levels of use related changes (Adams 1993). In the American southwest, Adams was able to draw the important conclusion that ‘morphological changes reflect the development of food-grinding technology’ and that a progressive improvement in productivity could be traced via the changes in shape of the formal grindstone sets (metates or basal millstones and their associated handstones or manos) (Adams 1993:331; also Bartlett 1933:10). Despite the ongoing debate about formal and amorphous grindstones, such distinctions are not possible in Australia where morphology seems to have been as heavily influenced by availability of raw

29 material as by any concerns about productivity (Hayes 2015; Smith 2015b). Formal grindstones have been identified in Australia but only in gross terms (Smith 1985) and little was understood about their relative efficiency other than an assumption that large millstones outperform generalised grindstones. A ‘variety of different implements are subsumed under the term ‘grindstone’’ (Smith 1985:23) and the complete array of metates discussed by Adams would, in Australia, be encompassed by the single term ‘millstone’. No comparable examination to that of Adams has been attempted in Australia.

If an evolution of grindstone technology occurred in Australia, and the changes through time are to be understood, the possibly subtle differences between the various implement attributes will need to be established and myths pertaining to the similarity of Australian grindstones dismissed.

Human Behavioural Ecology and Productivity Australian grinding studies have regularly provided data for Human Behavioural Ecology (HBE) models (see Bird and Bliege Bird: 2005 for a general background). Except in very remote areas, by the 1970s, grass seed had long been supplanted as a staple by European foods including white flour (O’Connell and Hawkes 1981; O’Connell et al. 1983). The movement away from seeds whilst other bush foods were still being gathered led O’Connell and Hawkes to suggest that it was mainly the high ‘cost’ of the seeds - gathering, cleaning and grinding and also the acquisition of processing technology (1981:110) - that ranked seeds lower in an optimal foraging model than many other available bush foods. Their work became a cornerstone of HBE in Australia (Bird and Bliege Bird 2005). Seed-grinding constituted a significant component in their model. However, as discussed later, some of their assumptions about the labour costs of the actual grinding of seeds are based on limited evidence (Hawkes and O’Connell 1981; O’Connell and Hawkes 1981).

The productive efficiency of various types of Australian seed-grinding stones is not well understood. Large, formal grindstones were expensive in terms of provisioning costs, that is, the costs of raw material selection and acquisition, transportation, security and so on. Their economic justification is difficult to understand if their productive performance is undifferentiated from that of smaller, convenient, less costly grindstones as there is a constant trade-off between efficiency and technological costs (Bright et al. 2002). There is a general acceptance that large grindstones should perform more efficiently than smaller examples (e.g. Adams 1993, 2002; Bright et al. 2002; Clarkson and Shipton 2015), but specific supporting ethnographic or empirical evidence is scarce. An exception is the work of Hard and Mauldin who demonstrated a positive relationship between an increase in the size of manos (top stones) and production (Hard et al.1996; Mauldin 1993). These results need to be verified and expanded to various classes of basal grindstones as well as topstones.

Use of Replicative Studies The extent and success of replicative studies in evaluating the problems associated with grindstones, especially productivity or efficiency, has not been reported in detail.

30 Numerous calls for exploration of grinding technology by way of replication have been issued in Australia and elsewhere (e.g. Adams 1989a, 1999, 2002, 2010, 2014; Fullagar 2006; Hard et al. 1996; Veth et al. 1997). Nevertheless, despite grindstones being considered ‘well-suited to the application of a variety of experimental and analytical techniques’ (Rowan and Ebeling 2008:6), responses in readily available English language publications have been limited. One possible reason is that, despite such ‘suitability’, the numerous and wide-ranging variations in grindstone and handstone raw material and morphology that were acceptable to traditional makers (Hayden 1987:31) and the range of materials ground ensures that comparative data are limited and difficult to apply (David 1998). The wide differences in the morphology of seeds ground by Aboriginal Australians is illustrated in Figure 2. As such, experiments that are consistently specified and designed to serve as ‘building blocks’ which will integrate to build a picture of the role of grindstones in the economy being studied, are needed.

A. B. C. Figure 2. Examples of seeds with wide morphological differences ground by Aboriginal Australians. Background grid is 2 mm. A. Portulaca oleracea (pigweed), B. Acacia aneura (mulga) and C. Brachtchiton gregorii (desert kurrajong).

A number of experimentalists have utilised some form of replication to examine aspects of grindstone technology (e.g. Fullagar et al. 2008: 2012: 2015; Hayes 2015). For example, Hayes reproduced 26 grindstones in investigating use-wear and residues on sandstone of varying grades of hardness, but did not report on their production (2015:148-150). Until recently, only Adams (1989a and b, 1999), Hard et al. (1996), Mauldin (1993) and Wright (1993) considered the productive efficiency of grindstones.

Adams, who has undertaken extensive ground-stone research over the past two decades, has mainly concentrated on use-wear issues (with a secondary focus on motor habits) and her production experiments were ‘designed to evaluate which stone characteristics were important in the formation of wear’ (2014:134). As such, actual measured production was not usually critical and she determined output, by volume, in terms of ‘cups’ (presumably an American Standard Cup equivalent to 236 ml). In summary, she found that a cup of a number of substances — corn, amaranth, clay and sherds — could be ground to flour consistency in 15 to 20 minutes (Adams1989a, 1993, 2010). As she measured by volume only to quarter cups and used mainly corn, a grain with no realistic Australian native proxy, relevance to Australian conditions is only indirect.

Both Hard et al. and Mauldin applied modern industrial comminution modelling science to investigate the effects of changes in the grinding surface area of manos

31 and metates as indicators of agricultural intensification. They recorded times to grind a kilogram of corn using manos with increasing grinding surface areas (Hard et al. 1996:257; Mauldin 1989:319) and concluded that, since grinding times decreased as grinding surfaces increased, ‘a strong relationship existed between grinding area and grinding rates’ (Hard et al. 1996:256). However, because of the limited information on the grindstones involved and the use of corn, direct comparisons with Australian studies are again difficult.

Wright measured the rates of attrition and production of the grindstone pair during prolonged grinding of corn. She reported production of between 624 and 864 grams of meal per grinding hour using three different handstones. However, again she utilised corn and a style of millstone (troughed metate) not known in Australia (Wright 1993). Simms, in the Great Basin of the United States, collected and processed numerous seeds to determine their ranking in foraging models but generally only to the stage where they were suitable for storage. He considered that the extra step of grinding the seeds would have added little to handling times (1984:80). Nevertheless, he does provide one grinding example and states that 62 grams of Indian Rice Grass ‘could easily be ground on a metate in 2 minutes’ (1984:85). It is unclear whether this was an actual experiment or merely an estimate but regardless, does provide one indication that the grinding of small seeds may not have always involved the high labour costs usually ascribed to the task.

In Australia, in addition to those mentioned above, Chris Clarkson and his students have performed basic production experiments at the University of Queensland (e.g. Clarkson and Shipton 2015). O’Connell et al. (1983) and Smith et al. (2015a & b) each mention a small seed-grinding experiment but few details are provided. These are the only reports of experimental seed-grinding production rates in Australia published to date but are now expanded and supplemented by experiments performed by Mildwaters and Clarkson (2018) who replicated 11 grindstones of varying morphology and, using three proxy commercial grains, investigated a number of aspects of grindstone productivity as discussed later. Figure 3 shows one of the replicated millstones being used by the author to grind Acacia aneura.

Figure 3. Author grinding parched Acacia aneura (mulga) seeds using a replicated sandstone millstone and pebble muller

32 Ethnographic Grinding Studies The available literature provides a small number of examples of timed ethnographic grinding sessions but, on examination, most have problems when assessing grinding performance. An example will illustrate the difficulties inherent in applying even the limited information available. Smith, in his widely quoted 1986 paper, states that a processing time of from two to five hours is necessary to produce a kilogram of flour based on the times given in a number of other papers (discussed below) which focused on the amount of seed ground - rather than flour produced (1986:31). The wide spread of two to five hours itself makes detailed analyses problematic as the period could represent a relatively minor task, or a large portion of a day’s labour. In addition, these production rates are far from unequivocal as it is noted that the times are reduced to two to four hours in a later paper (Smith 1989:313) but then increased to two to six hours (Smith 2013:198). No explanation for the spread of times is provided but it may be that it is intended to account for such matters as variations in grindstone size and seed type. As well, other problems inherent in the underlying research further complicate application of the data.

Brokensha (1975:25) commissioned women who, by that time, normally used white flour, to ‘prepare traditional damper’ from grass seeds. He gives times for collection of two kilograms of rough seed (4.5 h/kg) but does not state the amount of clean seed available for grinding. A combined grinding and cooking time of two hours produced 1800 ml - or 900 ml/h - of baked damper, not ground meal. No precise time to produce ground meal is offered but Smith, using available evidence, has calculated a likely grinding rate (see later) of about eight hours per kilogram of seed (2015b:117).

O’Connell and Hawkes observed a brief: demonstration of grinding [of Acacia aneura which] indicates est. rate of 165 g/hr or 6.06 hr/kg. P. Latz (personal communication) says this figure may be too high. His experiments suggest a grinding rate of 3 hr/kg. Combining these data with our own, we estimate [grinding time] (including parching) [parching only 3.6 minutes/kg] = 4 hr/kg’ (1981: Notes 124-5).

O’Connell and Hawkes’ processing times for a number of different seeds consist of the recorded collection times plus the above calculated four hours for grinding. The grinding time is thus only an averaged approximation based on limited personal observation and a brief comment from a colleague. The calculated grinding time has been absorbed into later works as a measured time.

O’Connell (et al. 1983) appear to use very similar, possibly the same, data but in this case, Latz actually ground ‘50 g of Acacia holosericea seed (slightly smaller than mulga) in just under 13 min, which suggests a rate of 4 h/kg [actual 4.34]’ (1983:92). No details of the equipment used are provided but a grooved millstone is illustrated. O’Connell’s ethnographic observation suggesting a rate of 6.06 h/kg is abandoned and Latz’s data from a brief 13 minute experiment is adopted. As mentioned earlier, high level theory building has been influenced by these works.

Only the final researcher, Cane, undertook detailed and well controlled experiments and was able to compile ‘the only quantified data of this kind in Australia’ (1984:77). Cane meticulously detailed individual times for processing steps from collection to

33 cooking for a number of different seeds, all soft grasses or herbs, including dividing some samples into batches (his Tables 4.3 to 4.7, 1984:77-80). Some of the seeds were pre-soaked before grinding. A large grooved grinding slab, or millstone, was used on all occasions. This fine level of detail reveals some of the problems that can arise when relying on brief demonstrations. From Table 4.3, cleaning resulted in 485 g of raw Panicum australiense and Fimbristylis oxystachya seed (often found together) of which 1/5 or 97 g was ground in 41 minutes; a production rate of 7.05 hours per kilogram. The remaining 4/5 or 388 g was ground in 46 minutes at a vastly more productive rate of 1.98 hours per kilogram. The average calculated grinding rate of 3.01 hours per kilogram is considerably less than Cane’s own estimate of ‘about one hour to grind approximately 200 g’ (or five hours per kilogram) (Cane 1984:78). These averages thus both emphasise and disguise a wide spread of times. Although all grinding was of small, soft seeds there would appear to be clear differences in the grinding skills or in the effort required to grind the seeds. Whilst it is becoming difficult to test seed-grinding skills due to aging of those with traditional experience, the grinding characteristics of various seeds, including the need to parch or soak before grinding, can still be experimentally explored. In 2015, Smith - citing Cane 1984 and O’Connell and Hawkes 1981- suggested that a kilogram of cereal seed could be ground in four hours ‘by women using millstones’ but that this rate needed to be confirmed (2015b:118). As noted above, Cane’s ‘estimated rate’ was five hours and O‘Connell and Hawkes’ calculated time of four hours was based on grinding an acacia, not a cereal.

In a field experiment mentioned earlier, Smith et al. arranged for 126 g of Panicum (Yakirra australiense) to be ground by two elderly women. The seed was dry ground on an unmodified quartzite slab - a tjiwa, not a millstone - in 60 minutes; a rate of almost eight hours per kilogram (2015:78). The weight of the flour produced was not reported.

Whilst discussing general purpose tjiwa grindstones, Smith appears to utilise this same experiment to help quantify the output of Brokensha’s earlier mentioned report (2015b:117). Brokensha’s data are used to support various other propositions, in this case that ‘ethnographic seed-grinding averaged 2 hours/day (to produce 500 g of flour)’ at a grinding rate of four hours per kilogram (Smith et al. 2015:79; Smith 2015b:117). The various investigations described above are summarised in Table 1.

It is apparent that reported grinding rates can vary widely and that there are elements of both repetition and circularity in some of the reports and data. Accordingly, calculations based on any particular observation may possibly contain significant disparities.

34 Researcher Reference Grindstone Area Seed Name Wet / Dry / Treatment Grind Produce Comments Not Stated Seed Meal (including whether Estimate or Actual) (N.S.) (Hours) (Hours) Brokensha 1975:25-7 Flat Tomkinson Panicum Wet None 8.00 Estimate. 900 cc of damper produced in 1 hour. sandstone Ranges, SA, decompositum-native Grinding rate /hour unclear. Smith's (2015:117) grinding NT, WA corner, millet estimate of 125g/hour seems most likely. slab-tjiwa. Western Desert. O'Connell & 1981:124-5 Flat or Sandover Acacia aneura- N.S. (Dry?) Parched 6.06 Actual. 'Brief demonstration of grinding indicates Hawkes grooved slab River, NE of mulga est. rate of 165 g/h or 6.06 hr/kg' (124) Alice Springs, NT. O'Connell & 1981:124-5 Flat or Sandover Acacia aneura- N.S. (Dry?) Parched 3.00 Actual. 'P. Latz (personal communication) says Hawkes grooved slab River, NE of mulga this figure may be too high. His experiments Alice Springs, suggest a grinding rate of 3 hr/kg' (124). NT. O'Connell & 1981:124-5 Flat or Sandover Acacia aneura- N.S. (Dry?) Parched 4.00 Actual. 'Combining these data with our own, we Hawkes grooved slab River, NE of mulga estimate ... (including parching) = 4 hr/kg' less Alice Springs, parching 0.06 = 3.94 h/kg. NT. O'Connell et 1983:92 Grooved Sandover Acacia aneura- N.S. (Dry?) Parched 6.06 Actual. 'A brief demonstration of grinding indicated al. grinding slab River, NE of mulga est. rate of 165 g/h or about 6 hr/kg' (92). Same Alice Springs, experiment as O'Connell & Hawkes 1981. NT. O'Connell et 1983:92 Grooved Sandover Acacia holosericea- N.S. (Dry?) Parched 4.34 Actual. ' Latz thinks this figure may be too high; he al. grinding slab River, NE of candelabra wattle ground 50 g of Acacia holosericea seed (slightly Alice Springs, smaller than mulga) in just under 13 min, which NT. suggests a rate of 4 h/kg' (92). Same experiment as above. Cane 1984:77-84 Large Western P. australiense- Wet None 7.05 Actual. 20% sample of 485g or 97g untreated & 1987:401- grooved Desert, WA. Flinders grass and F. seed ground in 41 minutes. 3 oxystachya- fringe rush Cane 1984:77-84 Large Western As above Wet None 1.98 Actual. 80% sample of 485g or 388g untreated & 1987:401- grooved Desert, WA. seed ground in 46 minutes. 3 Cane 1984:77-84 Large Western Fimbristylis Wet None 1.48 Actual. 2/3 sample of 340g or 226g untreated & 1987:401- grooved Desert, WA. oxystachya-fringe seed ground in 20 minutes. 3 rush Cane 1984:77-84 Large Western Fimbristylis Wet None 2.20 Actual. 1/3 sample of 340g or 114g untreated & 1987:401- grooved Desert, WA. oxystachya-fringe seed ground in 15 minutes by a young girl. 3 rush Cane 1984:77-84 Large Western Panicum Wet Soaked 3.65 Actual. 160g seed soaked for 2h 17minutes & 1987:401- grooved Desert, WA. cymbiforme- ground in 35 minutes. 3 gumbulyu Cane 1984:77-84 Large Western Chenopodium Wet Soaked 2.76 Actual. 255g seed soaked for 8 minutes ground in & 1987:401- grooved Desert, WA. rhadinostachyum 42 minutes. 3 35 Smith 1986:31 Not stated Australian Arid Various grass or soft Wet None 2.00 to Estimate. From Brokensha 1975, O'Connell & Zone seeds 5.00 Hawkes 1981, O'Connell et al 1983, Cane 1984 to produce 1 kg flour. Smith 1986:31 Not stated Australian Arid Various acacia or Wet Parched 2.00 to Estimate. From Brokensha 1975, O'Connell & Zone hard seeds 5.00 Hawkes 1981, O'Connell et al 1983, Cane 1984 to produce 1 kg flour. Smith 1986:31 Not stated Australian Arid Very hard Acacaia Wet Parched & 2.00 to Estimate. From Brokensha 1975, O'Connell & Zone victoriae, A. coriacea Crushed 5.00 Hawkes 1981, O'Connell et al 1983, Cane 1984 to and others produce 1 kg flour. Smith 1989:313 Not stated Australian Arid Various grass or Wet N.S. 2.00 to Estimate. 'Estimates 'of the processing time Zone acacia seeds 4.00 required to produce a kilogram of flour from various grass or acacia seeds range from two to four hours' (313) based on calculations from Brokensha 1975 etc. Smith 2013:198 Millstone- Central Various seeds Wet N.S. 2.00 to Estimate. It 'takes 2-6 hours to produce a Large flat- Australia 6.00 kilogram of flour from most seeds' (198) based on surfaced slab Desert Areas Brokensha 1975, O'Connell & Hawkes 1981, O'Connell et al 1983, Cane 1987. Smith 2015:117 Quartzite Mann- Yakirra australiense- Dry None 7.94 Actual. 'two ... women, both experienced in tjiwa Mulgrave panicum australiense grinding seed, and working as a tag team, ground Ranges, 126 grams of … seed in one hour' (117). May be southern same experiment as Smith et al. 2015:78-9. Western Desert, SA. Smith 2015:117 Millstones Various arid Various seeds N.S. N.S. 4.00 Actual. From Cane and O'Connell & Hawkes - areas. Cane's times ranged from 1.48 to 7.05 h/kg. O'Connell & Hawkes was a calculated estimate. 'This compares with … (250 g per hour) by women using millstones' (117). Smith et al. 2015:78-9 Unmodified Mann Ranges, Yakirra australiense Dry None 7.94 Actual. 'In relay, these women ground 126 g of quartzite slab SA seed in 60 minutes' (78). May be the same experiment as Smith 2015: 117. A second similar experiment mentioned but no details provided. Table 1 : Summary of Grinding Production – Actual and Estimated

36 Discussion The effects of morphological characteristics, other than the earlier mentioned ‘big is better’ presumption, are far from agreed. For example, Smith suggests that for grooved millstones, because of ‘the long grinding action with which these implements are used’, those with a groove less than 300 mm in length (a far from small millstone) ‘are possibly too small to use efficiently’ (1985:26; 1986:32). Veth and O’Connor considered much smaller grindstones with only incipient grooves to have been centrally involved in seed-grinding (1996:20). Davidson and McCarthy believed that the grooved millstone was ‘the most specialized type’ and presumably possessed advantages over the ‘commonest variety [with] a symmetrical all-over [dish] depression’ or the mortar which occurred throughout Australia (1957:438-44; McCarthy 1976:63). Smith states that mortars were used in Central Australia ‘to pound very hard-coated seeds (mainly acacias) into a coarse meal, as a preliminary stage of processing before wet milling’ (2013:199). This suggests that mortars were not capable of producing the relatively fine meal or flour used for seed cakes.

Mildwaters and Clarkson (2018) addressed a number of such issues. They found that an increase in size of the grinding surface was the most influential factor affecting productivity but millstones smaller than the 300 mm discussed by Smith still performed well within the parameters suggested in the ethnographic literature. Whilst the very small grindstones such as those described by Veth and O’Connor ‘have noteworthy grinding capability’, Mildwaters and Clarkson suggest it is unlikely that they would have been utilised as specialised seed-grinders (2018:17). The type of seed, and whether ground wet or dry, were also found to be of major importance with wet grinding providing a productive advantage with each type and size of grindstone. A mortar was one of the implements reproduced by Mildwaters and Clarkson, and whilst they were unable ‘to picture the mortar as a primary seed processing implement’, they found that it was able to produce adequately fine meal at a rate well within the range of the ethnographic observations (2018:14).

What then can be said about the ethnographic records? O’Connell and Hawkes’ study suggested that the high costs of seed use, of which grinding constituted around half (Cane 1984), encouraged the consumption of alternative resources. This assessment was based on an estimated time of four hours to grind one kilogram of seed on (judging from other comments and an illustration) a large grooved millstone (O’Connell and Hawkes 1981:124). However, only one of Cane’s experiments using a large grooved millstone required this length of grinding time, namely a small initial sample of 97 g (out of a total 1240 g) ground at a rate 7.05 h/kg. The other five grindings required from 1.48 to 3.65 h/kg, an average of 2.33 h/kg. Half of Cane’s experiments required an average grinding time of only 1.89 h/kg. Support for the general applicability of Cane’s results is provided by the experiments of Mildwaters and Clarkson (2018).

Their experiments with a large millstone bearing artificially produced grooves, averaged 1.6 hours to produce a kilogram of wet meal (their Table 4). However, the amount of seed ground and the amount of meal produced, especially with wet grinding, can be quite different due to various factors including the absorption of moisture in the meal. Accordingly, data from their experiments have been re-

37 formatted to include the time taken to grind a kilogram of seed under various constraints (see Table 2).

Produce 1 Use 1 kg kg Grindstone Seed (Common Names) Seed Wet Meal (hours) (hours) French Millet, Green Panic, Red Large Grooved Used - 1 1.6 2.7 Sorghum Large Grooved Used - 1 French Millet 1.2 1.5 Large Grooved Used - 1 Green Panic 2.1 4.7 Large Grooved Used - 1 French Millet & Green Panic 1.6 2.8 French Millet, Green Panic, Red All Small - 3; 4; 5; 8; 9 4.1 5.9 Sorghum All Large - 1; 2; 6; 7; 10; French Millet, Green Panic, Red 1.8 3.1 11 Sorghum French Millet, Green Panic, Red All - 11 grindstones 2.9 4.4 Sorghum Table 2 - Grinding Hours to produce 1 kg of Wet Meal or use 1 kg of various seeds

Using the Large Grooved Used millstone (LGU 1), average time to grind a kilogram of seed with all three proxy seeds was 2.7 hours compared with 1.6 hours to produce an equivalent weight of wet meal. Excluding Cane’s initial small sample, the times are comparable. It is important to note that here, like is being compared with like - large grooved millstones and small easy to grind seeds were used in both instances. However, a further discrimination is possible by assessing the two experimental grains individually. A kilogram of the sharp, brisant millet was ground in 1.5 hours but the soft, spongy panic required 4.7 hours demonstrating, as Cane found, that seemingly similar seeds can have very different grinding profiles. The only available well controlled large ethnographic study and the only detailed experiment thus both suggest that at least some seeds can be ground on a large grooved millstone in much less time than that utilised by O’Connell and Hawkes in their calculations. As such, the pressure away from seeds towards alternative resources may not always have been as intense as proposed. Similarly, estimates of the extent of female labour devoted to grinding in seed-based economies may need to be reassessed.

As discussed earlier, Smith suggested that specialised millstones were around twice as efficient as general purpose tjiwa grindstones in processing cereal seeds. That is, they required an average of only around four hours to grind a kilogram of seed as opposed to around eight hours for the tjiwa. Mildwaters and Clarkson’s group of large experimental millstones were able, on average, to grind a kilogram of seed in 3.1 hours, (see their Table 2, 2018; re-formatted data), which is in reasonable alignment with Smith’s (2015b) estimate. However, when the group of smaller grindstones, which included small, non-specialised grindstones and a mortar, were considered, the time increased materially to 5.9 hours. A replicated tjiwa was not part of Mildwaters and Clarkson’s experiments so it is not possible to comment on the specific productivity of that implement. In general, these results tend to confirm that a rate of four hours grinding time for a kilogram of seed using a millstone appears reasonable (Smith 2015b) when used in routine, non-critical contexts. However, when important productivity or efficiency issues are involved it is important to make

38 clear that, with the large ‘specialised’ millstones and ‘grass’ type seeds, substantially lower times were possible. Regarding production rates, it is interesting to note that in ancient Mesopotamia, records dating to about 5,000 years BP from large scale milling complexes suggest specialist women millers were able to grind about two and a quarter kilograms of barley flour each per day; a probable rate of around four hours per kilogram (Gregoire 1999) which agrees well with the Australian experience.

Conclusion The foregoing assessment of fundamental grinding problems clearly demonstrates that many of the basic technological issues associated with using a millstone for food grinding are still not well understood. As a consequence, both practical and theoretical assessments depending on grindstone efficiency must, of necessity, be relatively coarse-grained.

Few ethnographic studies are now possible with women who ‘ground for a living’ as distinct from those who may have merely observed, or played at, grinding in their youth. However, the time frame over which the economic contribution of grindstones needs to be considered has now extended to 65,000 years. It thus seems likely that more and more reliance will need to be placed on experimental studies.

Valid questions can be raised about the applicability of experimental studies and the skills of experimental grinders but questions need also be raised about ethnographic studies and the proficiency of millers returning to country and demonstrating seed- grinding after relying on white flour for up to forty years. Grinding is a skilled manual endeavour and, like any such task, needs to be practiced if skills are to be gained and maintained. Similarly, ideal workshop or laboratory conditions cannot be directly compared to those existing in traditional camps where various distractions almost certainly impinged on grinding efficiency. Whilst experimental results cannot necessarily be translated directly back to traditional lifestyles, they do provide important bases for comparison. Within these limitations, Mildwaters and Clarkson have demonstrated that many of the baseline technological factors in question are able to be effectively examined by appropriately designed experiments.

However, experiments to date barely touch on what could be done and what is needed. Mildwaters and Clarkson state they are at present investigating the grinding characteristics of native seeds, and that their research on the effects of the topstone of the grinding pair and the comparison of reciprocal versus rotary grinding actions when grinding surface areas are similar is continuing (pers. comm Clarkson 2018). Nonetheless, much more needs to be done. For example, the effects of the raw material type used in the grindstone needs to be established. Also the potential of possible grindstone types not discussed above (other than those usually associated with central Australian seed areas) such as tjiwa grindstones (Smith 2015b), axe- grinding grooves and dimpled anvils (Pardoe 2015) should be evaluated.

The debates and theories discussed earlier would often take on improved clarity if the breadth, capability and development of the underlying technologies were explicated. For example, the grinding stone ‘sameness’ assumptions and the ‘formal’ debate have possibly narrowed thinking about the land areas wherein seed-grinding

39 may have been a contributor of economic importance and also obscured possibly widely varying female labour inputs.

It is expected that, with a better appreciation of grinding basics, both ethnographic and archaeological examples should be capable of providing more instructive information on the traditional economies of Aboriginal people. Additional well- controlled experiments would provide a worthwhile portion of the missing information. It is hoped that the availability of such basic technological building- blocks might encourage a visit, or re-visit, to seed-grinding by both archaeologists and those in related disciplines.

40 References Adams, J.L. 1989a. ‘Experimental replication of the use of ground stone tools’. Kiva 54(3):261-271. Adams, J.L. 1989b. ‘Methods for improving ground stone artifacts analysis: experiments in mano wear patterns’. In D.S. Amick and R.P. Mauldin (eds), Experiments in Lithic Technology, pp. 259-276. Oxford: BAR International. Adams, J.L. 1993. ‘Toward understanding the technological development of manos and metates’. Kiva 58(3):331-344. Adams, J.L. 1994. The Development of Prehistoric Grinding Technology in the Point of Pines Area, East-Central Arizona. Ann Arbor: UMI Dissertation Services. Adams, J.L. 1999. ‘Refocusing the role of food-grinding tools as correlates for subsistence strategies in the U.S. Southwest’. American Antiquity 64(3):475- 498. Adams, J.L. 2002. Ground Stone Analysis: A Technological Approach. Salt Lake City: University of Utah Press. Adams, J.L. 2010. ‘Understanding grinding technology through experimentation’. In J.R. Ferguson (ed.), Designing Experimental Research in Archaeology: Examining Technology Through Production, pp. 129-151. Boulder: University Press of Colorado. Adams, J.L. 2014. ‘Ground stone use-wear analysis: a review of terminology and experimental methods’. Journal of Archaeological Science 48:129-138. Allen, H. 1998. ‘Reinterpreting the 1969-1972 Willandra Lakes Archaeological Surveys’. Archaeology in Oceania 33(3):207-220. Balme, J. 1991. ‘The antiquity of grinding stones in semi-arid western New South Wales’. Australian Archaeology (32):2-9. Bartlett, K. 1933. ‘Pueblo milling stones of the Flagstaff Region and their relation to others in the Southwest’. Museum of Northern Arizona Bulletin (No.3):5-32. Beck, W., A. Clarke and L. Head 1989. ‘Plants in hunter-gatherer archaeology’. Tempus 1:1-13. Binford, L.R. 1979. ‘Organization and formation processes: looking at curated technologies’. Journal of Anthropological Research 35(3):255-273. Bird, D.W. and R. Bliege Bird 2005. ‘Evolutionary and ecological understandings of the economics of desert societies: comparing the Great Basin USA and the Australian Deserts’. In P. Veth, M. Smith and P. Hiscock (eds), Desert Peoples: Archaeological Perspectives, pp. 81-99. Malden, Mass.: Blackwell. Bright, J., A. Ugan and L. Hunsaker 2002. ‘The effect of handling time on subsistence technology’. World Archaeology 34(1):164-181. Brokensha, P. 1975. The Pitjantjatjara and Their Crafts. Sydney: Aboriginal Arts Board, Australia Council. Cane, S. 1984. Desert camps: a case study of stone artefacts and Aboriginal behaviour in the Western Desert. Unpublished PhD thesis, The Australian National University, Canberra. Cane, S. 1987. ‘Australian Aboriginal subsistence in the Western Desert’. Human Ecology 15(4):391-434. Clarkson, C. and C. Shipton 2015. ‘Teaching ancient technology using “hands-on” learning and experimental archaeology’. Ethnoarchaeology 7(2):157-172. Clarkson, C., Z. Jacobs, B. Marwick, R. Fullagar, L. Wallis, M. Smith, R. G. Roberts, et al. 2017. ‘Human occupation of northern Australia by 65,000 years ago’. Nature. 547 pp. 306-310.

41 David, N. 1998. ‘The ethnoarchaeology and field archaeology of grinding at Sukur, Adamawa State, Nigeria’. The African Archaeological Review 15(1):13-63. Davidson, D.S. and F.D. McCarthy 1957. ‘The distribution and chronology of some important types of stone implements in Western Australia’. Anthropos 52(3 & 4):390-458. deBeaune, S.A. 2004. ‘The invention of technology: prehistory and cognition’. Current Anthropology 45(2):139-162. Devitt, J. 1992. ‘Acacias: a traditional Aboriginal food source in central Australia’. In A.P.N. House and C.E. Harwood (eds), Australian Dry-zone Acacias for Human Food: Proceedings of a Workshop held at Glen Helen, Northern Territory, Australia, pp. 37-53. Melbourne: CSIRO Publications. Dix, W.C. and M.E. Lofgren 1974. ‘Kurumi: possible Aboriginal incipient agriculture associated with a stone arrangement’. Records of the Western Australian Museum 3(1):73-77. Dubreuil, L. 2004. ‘Long-term trends in Natufian subsistence: a use-wear analysis of ground stone tools’. Journal of Archaeological Science 31(11):1613-1629. Edwards, D.A. and J.F. O’Connell 1995. ‘Broad spectrum diets in arid Australia’. Antiquity (69):769-83. Fullagar, R. 2006. ‘Starch on artifacts’. In R. Torrence and H. Barton (eds), Ancient Starch Research., pp. 177-203. Walnut Creek, Calif.: Left Coast Press. Fullagar, R., J. Field and L. Kealhofer 2008. ‘Grinding stones and seeds of change: starch and phytoliths as evidence of plant food processing’. In Y.M. Rowan and J.R. Ebeling (eds), New Approaches to Old Stones: Recent Studies of Ground Stone Artifacts, pp. 159-172. London: Equinox. Fullagar, R., L. Liu, S. Bestel, D. Jones, W. Ge, A. Wilson & S. Zhai. 2012. ‘Stone tool-use experiments to determine the function of grinding stones and denticulate sickles’. Indo-Pacific Prehistory Association Bulletin. Vol.32, No. 1. pp. 29-44. Fullagar, R., E. Hayes, B. Stephenson, J. Field, C. Matheson, N. Stern and K. Fitzsimmons 2015. ‘Evidence for Pleistocene seed grinding at Lake Mungo, south-eastern Australia’. Archaeology in Oceania 50:3-19. Gorecki, P., M. Grant, S. O'Connor and P. Veth 1997. ‘The morphology, function and antiquity of Australian grinding implements’. Archaeology in Oceania 32(2):141-150. Gregoire, J.-P. 1999. ‘Major units for the transformation of grain: the grain-grinding households (e₂-HAR.HAR) of Southern Mesopotamia at the end of the Third Millennium BC’. In P.C. Anderson (ed.), Prehistory of Agriculture: New Experimental and Ethnographic Approaches, pp. 225-237. Los Angeles: University of California. Hard, R.J., R.P. Mauldin and G.R. Raymond 1996. ‘Mano size, stable carbon isotope ratios, and macrobotanical remains as multiple lines of evidence of maize dependence in the American Southwest’. Journal of Archaeological Method and Theory 3(3):253-318. Haury, E.W. 1950. The Stratigraphy and Archaeology of Ventana Cave, Arizona: University of Arizona Press. Hawkes, K. and J.F. O'Connell 1981. ‘Affluent hunters? Some comments in light of the Alyawara case’. American Anthropologist 83(3):622-626. Hayden, B. 1987. ‘Traditional metate manufacturing in Guatemala using chipped stone tools’. In B. Hayden (ed.), Lithic Studies Among the Contemporary Highland Maya, pp. 8-119. Tucson: The University of Arizona Press.

42 Hayes, A.C., J.N. Young and A.H. Warren 1981. ‘Excavation of Mound 7: Gran Quivira National Monument, New Mexico’. Publications in Archaeology 16. Hayes. Elspeth. 2015. What was ground? A functional analysis of grinding stones from Madjedbebe and Lake Mungo, Australia. Unpublished Ph.D. thesis. Wollongong: University of Wollongong. Johnson, J.E. 1963. ‘Observations on some Aboriginal campsites in South Australia and adjoining states’. Mankind 6(2):64-79. Johnson, J.E. 1964. ‘Observations on some Aboriginal campsites in South Australia and adjoining states’. Mankind 6(4):155-181. Mauldin, R. 1993. ‘The relationship between ground stone and agricultural intensification in western New Mexico’. Kiva 58(3):317-330. McCarthy, F.D. 1976. Australian Aboriginal Stone Implements: including bone, shell and teeth implements. Sydney: Australian Museum Trust. Mildwaters, J. and C. Clarkson. 2018. ‘The efficiency of Australian grindstones for processing seed: a quantitative experiment using reproduction implements and controlling for morphometric variation and grinding techniques’. Journal of Archaeological Science: Reports. 17, pp. 7-18. O’Connell, J.F. 1977. ‘Aspects of variation in central Australian lithic assemblages’. In R.V.S. Wright (ed.), Stone Tools as Cultural Markers: Change, Evolution and Complexity, 269-281. Canberra: Australian Institute of Aboriginal Studies. O’Connell, J.F. and K. Hawkes 1981. ‘Alyawara plant use and optimal foraging theory’. In B. Winterhalder and E.A. Smith (eds), Hunter-Gatherer Foraging Strategies: Ethnographic and Archaeological Analyses, pp. 99-125. Chicago: University of Chicago Press. O'Connell, J.F., K.L. Peter and P. Barnard 1983. ‘Traditional and modern plant use among the Alyawara of Central Australia’. Economic Botany 37(1):80-109. O’Connor, S., P. Veth and C. Campbell. 1998. Serpent’s Glen Rockshelter: report of the first Pleistocene aged occupation sequences from the Western Desert. Australian Archaeology. (46): 12-22. Pardoe, C. 2015. Grinding stones of the Lachlan River; archaeological studies along the Lachlan River; Condobolin, Lake Cargelligo, West Wyalong and Hillston regions. Part of the research project ‘Kiacatoo Man: biology, archaeology and environment at the Last Glacial Maximum. March 2015 Rowan, Y.M. and J.R. Ebeling 2008. ‘Introduction: the potential of ground stone studies’. In Y.M. Rowan and J.R. Ebeling (eds), New Approaches to Old Stones: Recent Studies of Ground Stone Artifacts, 1-15. London: Equinox. Schiffer, M.B. and J.M. Skibo 1987. ‘Theory and experiment in the study of technological change’. Current Anthropology 28(5):595-622. Simms, S.R. 1984. Aboriginal Great Basin foraging strategies: an evolutionary analysis. Unpublished PhD thesis, University of Utah, Ann Arbor. Simms, S.R. 1985. ‘Acquisition cost and nutritional data on Great Basin resources’. Journal of California and Great Basin Anthropology 7(1):117-126. Smith, M., E. Hayes and B. Stephenson 2015. ‘Mapping a millstone: the dynamics of use-wear and residues on a Central Australian seed-grinding implement’. Australian Archaeology 80:70-79. Smith, M.A. 1985. ‘A morphological comparison of Central Australian seedgrinding implements and Australian Pleistocene-age grindstones’. The Beagle, Occasional Papers of the Northern Territory Museum of Arts and Sciences 2(1):23-38.

43 Smith, M.A. 1986. ‘The antiquity of seedgrinding in arid Australia’. Archaeology in Oceania 21(1):29-39. Smith, M.A. 1989. ‘Seed gathering in inland Australia: current evidence from seed- grinders on the antiquity of the ethnohistorical pattern of exploitation’. In D.R. Harris and G.C. Hillman (eds), Foraging and Farming: The Evolution of Plant Exploitation, pp. 305-317. Sydney: Unwin Hyman. Smith, M.A. 1993. ‘Biogeography, human ecology and prehistory in the sandridge deserts’. Australian Archaeology (37):35-50. Smith, M.A. 2004. ‘The grindstone assemblage from Puritjarra rock shelter: investigating the history of seed-based economies in arid Australia’. In T. Murray (ed.), Archaeology from Australia, pp. 168-186. Melbourne: Australian Scholarly Publishing. Smith, M.A. 2013. The Archaeology of Australia's Deserts. Cambridge: Cambridge University Press. Smith, M. A. 2015a. Comment: What sort of seed grinding at Pleistocene Lake Mungo? Archaeology in Oceania. Vol. 50. pp. 175-176. Smith, M. 2015b. ‘Western Desert tjiwa-and tjungari- type grindstones and their archaeological significance’. Australian Aboriginal Studies. Vol. 1. pp. 115- 121. Stahl, A.B. 1989. ‘Plant-food processing: implications for dietary quality’. In D.R. Harris and G.C. Hillman (eds), Foraging and Farming: The Evolution of Plant Exploitation, pp. 171-194. Sydney: Unwin Hyman. Tindale, N.B. 1959. ‘Ecology of primitive Aboriginal man in Australia’. In A. Keast, R.L. Crocker and C.S. Christian (eds), Biogeography and Ecology in Australia, pp. 36-51. Den Haag: Junk. Tindale, N.B. 1977. ‘Adaptive significance of the panara or grass seed culture of Australia’. In R.V.S. Wright (ed.), Stone tools as Cultural Markers: Change, Evolution and Complexity, pp. 345-349. Canberra: A.I.A.S. Veth, P., R. Fullagar and R. Gould 1997. ‘Residue and use-wear analysis of grinding implements from Puntutjarpa Rockshelter in the Western Desert: current and proposed research’. Australian Archaeology (44):23-25. Veth, P. and S. O'Connor 1996. ‘A preliminary analysis of basal grindstones from the Carnarvon Range, Little Sandy Desert’. Australian Archaeology (43):20-22. Veth, P., M. Smith and M. Haley 2001. ‘Kaalpi: the archaeology of an outlying range in the dunefields of the Western Desert’. Australian Archaeology (52):9-17. Warner, W.L. 1937. A Black Civilization: A Social Study of an Australian Tribe. New York: Harper & Brother. Wright, M.K. 1993. ‘Simulated use of experimental maize grinding tools from southwestern Colorado’. Kiva 58(3):345-355.

3 Chapter 3 - Productive Efficiency of Various Grindstones

3.1 Introduction

The second paper examines the productive efficiency of various grindstones by way of experiments using a number of replicated grindstones and three proxy commercial seeds. It examines such matters as whether the morphology of the grindstone or the grinding action used influences the amount of meal produced or the amount of seed wasted and whether the proxy seeds used show significantly different production profiles.

3.2 Methodology

An experimental methodology was adopted. Ten basal millstones ranging in size from very small to large and with either differing functional surfaces or degrees of simulated wear, plus a mortar, were replicated in sandstone. Three proxy commercial seeds, Gatton green panic, White French millet and Queensland red sorghum, were selected for grinding in 253 sessions of ten minutes each. All inputs and outputs were electronically weighed.

3.3 Conclusion in Summary

In assessing efficiency of production, the area of the functional grinding surface is most critical. For grooved millstones of similar morphology, there is a more or less linear increase in efficiency as the length, and thus the area, of the groove increases. Regardless, small grindstones and mortars do have noteworthy seed grinding capability.

The seed being processed is also a critical factor in the output of ground meal. Based on the three domestic seeds used, seed selection may effectively double labour investment in grinding.

3.4 The Second Paper

The paper was published as: Mildwaters, John and Clarkson, Chris. 2018. The efficiency of Australian grindstones for processing seed: A quantitative experiment

using reproduction implements and controlling for morphometric variation and grinding techniques. Journal of Archaeological Science: Reports. Volume 17. pp. 7- 18. The paper, prior to final typesetting, is as follows:

The efficiency of Australian grindstones for processing seed: a quantitative experiment using reproduction implements and controlling for morphometric variation and grinding techniques Authors: John Mildwaters and Chris Clarkson

Abstract

This paper presents a controlled experimental examination of the efficiency of Australian Aboriginal grindstones with a variety of surface morphologies in milling seeds into meal. Several replicate sandstone grindstones (large and small millstones with functional surfaces ranging in length from 43 cm to 16 cm and a mortar) were employed to process three domesticated commercial grains that serve as viable proxies for native grains. These were processed in 10 minute grinding sessions. Our results show that large millstones significantly outperform both the small millstones and the small mortar in the net output of ground grain. We also find that other factors may influence productivity, including the amount of wear and the seed being processed. These variations are of sufficient magnitude to have possibly influenced crucial economic decisions.

Keywords Grinding; grindstones; seed processing; productivity; morphology; Australia; Aboriginal. Introduction Over past few decades a vigorous debate has emerged over the role played by grindstones in past Aboriginal lifeways; in particular, the antiquity of seed grinding, the importance of specialised tools (millstones), the role of seeds in the long-term exploitation of arid environments, the labour investment in seed collection and processing, and the extent to which seeds contributed to ancient diets (Gorecki et al. 1997; Smith 1985, 1986, 1989; 2013). Two opposing views have emerged in this debate; one which sees intensive seed grinding economies as a late Holocene development, and the other which traces seed grinding back to the Last Glacial Maximum (LGM) or before (e.g. Fullagar et al. 2008; 2015; Smith 1985; 1986; 1989; 2015a).While ethnographic, use-wear and residue evidence have all contributed significantly to this debate (e.g. Balme et al 2001; Fullagar 1991; Hayes 2015), a number of fundamental questions still remain about the role of grindstone form itself in this debate (Adams 2002). For example, what role does grindstone size, raw material type, muller type, presence or absence of a groove or type of seed processed play in milling efficiency, that is, the production of meal, labour inputs, grain wasted and so on? Are there significant differences in the efficiency of different types of grindstones when used to process the same or different seeds? If such performance characteristics make significant differences to grinding efficiency, then what might this mean for the evolution of grindstone form over time in Australia?

Addressing these questions may help understand whether intensified seed grinding at certain times in the past might have benefited from specialised equipment of some kind, thus resulting in the emergence of particular grindstone characteristics over time. This paper aims to provide a preliminary assessment of whether such fundamental issues as size, grinding action, muller type and variety of seed have significant effects on grindstone efficiency, and

47 to employ these low-level building blocks to reflect on the Australian grindstone debate. For this purpose, 280 controlled grinding trials were undertaken on 11 grindstones and three types of commercial seeds, plus tests on two native seeds, to determine the effects of form on efficiency. In particular we compare the efficiency of ‘formal’ millstones versus informal ‘amorphous’ grindstones and mortars, since these distinctions have been paramount in the Australian seed grinding debate.

Australian Seed grinding: Ethnography and Archaeology Australian archaeologists are fortunate to possess a rich and varied ethnographic record documenting the use of a variety of grindstones employed in a wide range of functions. Unfortunately, few of the ethnographic studies of seed grinding provide quantitative data on processing efficiency, either for different kinds of grindstones or seeds, though a number of more detailed studies do exist (see Cane 1984, 1987; Devitt 1992; Gould et al. 1971; O’Connell 1977; O’Connell et al. 1983). At present, much of our understanding of production and efficiency in arid seed grinding derives from O’Connell’s study of Alyawarra seed grinding. O'Connell worked with Central Australian Aboriginal women informants to gather and process a range of traditionally used native seeds. Quantitative data on processing times and outputs were recorded and his calculations of from four to six hours to grind a kilogram of seed have been widely accepted (1983; Bird and Bliege Bird 2005:90). While the Alyawarra work was a landmark study, few if any studies have since attempted to further quantify meal production rates on a wider range of grindstones or seeds in either an ethnoarchaeological or experimental setting.

A common distinction made in descriptions of Australian grindstones is between formal grooved millstones and informal amorphous or flat grindstones and mortars. However, what is meant by this is seldom clear. In his ethnographic work of 1897, Bennett distinguished between grooved (long, narrow depression) and basin (dish like depression) millstones. Other early ethnographers such as Spencer and Gillen provided brief descriptions of the grinding process but little on the grindstones themselves (1912), often merely referring to ‘the usual grinding stones’ (1899:22). Much later McCarthy (1976) and his colleagues provided what was accepted, until relatively recently, as the type description of many grindstones and pounders. However, it was not until the work of O’Connell (1977) and Smith (1985) that descriptions of the functional surface became the generally accepted means of classification. It is functional surface that is now used primarily to distinguish between grindstones in Australian archaeology. However, grindstone terminology remains confused and it is often difficult to be certain of the precise morphology of a grindstone under discussion.

The functional surface is a critical factor in the classification and efficiency of a millstone (O’Connell 1977; Smith 1985). Mauldin found the grinding surface area to be positively related to output. He considered the functional surface area of the topstone to be the best indicator of efficiency for the mill sets studied (1993:319). A further indicator of efficiency may be the degree of use-wear. Some researchers consider that unused or lightly used grindstones possess a productive advantage over well-worn examples (e.g. Gorecki et al. 1997; Veth and O’Connor 1996). However, whilst Gorecki et al. are clear that grindstones are ‘far more efficient and versatile in earlier stages of reduction than when worn’ (1997:142), an earlier observation by Warner suggests that a grinding stone which has developed a cavity ‘is sought for by a woman in preference to a new stone’ (1937:497). Groove development or absence also impinges on the question of formal or amorphous grindstones. Accepting for the moment that formal grindstones have well defined grooves but the surfaces of amorphous implements are flat with perhaps minor signs of abrasion (Veth and O’Connor 1996; Smith 1985), then the question of whether any groove is functional or merely the product of use wear is important.

From the 1940s to the 1970s, McCarthy and Tindale published a number of papers dealing with seed grinding technology and its distribution (e.g. McCarthy 1976; et al. 1946; Tindale 1959; 1977). McCarthy limited his focus to the definition of a number of grindstone types and produced the first widely adopted typology of Australian implements (et al. 1946; 1976). Tindale, on the other hand, proposed an evolutionary sequence of grassland use and grindstone development. He suggested that hammers and anvils were in use from the terminal Pleistocene and that ‘minor abrasional surfaces’ appeared from about 8,000 years BP. From around 4,000 years BP:

several styles of upper and nether millstones, including ones such as were used in the wet grinding of grass seed food, appear, and culminate in the grass-seed-meal preparing mill sets characteristic of [ethnographic times] (1959:49).

Morphological details of the early millstone sets are not provided. Tindale suggested that the use of millstone sets in late prehistory identified a populous, socially advanced dry areas grass-seed culture he called the Panara (1977).

In an influential paper, Smith (1985) proposed an updated version of McCarthy’s grindstone typology which divided Australian grindstones into five types on the basis of functional surface. These were: millstones, mullers, mortars, pestles and amorphous grindstones. The first four types were considered formal seed grinding implements incorporating elements of intentional design. The amorphous class, on the other hand, consisted of unmodified natural stones or slabs used expediently and lacking intentional design. He later expanded and clarified his view on amorphous grindstones as used in other areas lacking in sandstone raw materials (Smith 2015b). In Smith’s area mortars and pestles were used to dry crush hard seeds such as those of the Acacias before they were wet ground using the millstones and mullers in the same way as soft grass seeds (Smith 1985:24-29). For seed grinding millstones, he considered size to be critical as a ‘large surface area is necessary for the long grinding action with which these implements are used and specimens smaller than 300 x 400 mm are possibly too small to use efficiently’ (1985:26). However, Veth and O’Connor examined a large sample of grindstones from the Little Sandy Desert of Western Australia and judged them to be seed-grinders despite the maximum dimension of the largest being only 221 mm (1996:21).

A central tenet of the Smith paper was that seedgrinders could be distinguished from amorphous grindstones by their diagnostic morphological characteristics. A second tenet was that Pleistocene dates for seedgrinding could not be supported on archaeological evidence (1985:29 & 36). Smith later argued that seedgrinding was a mid to late Holocene ‘further development of an existing technology arising out of a need to more heavily exploit certain resources’ (Smith 1986:37). Thus Smith linked the emergence of new specialised grindstone morphology with an intensive economic reliance on seeds, implying that the new specialised technology increased the efficiency of this subsistence activity.

Smith’s arguments have since received widespread support (e.g. Balme 1991; Mulvaney 1998). However, not all agree with his typology, nor with the conclusions drawn from the archaeological evidence. For example, Gorecki et al. (1997) countered that the specialised seedgrinding toolkit identified by Smith may only be one component of a more generalised Pleistocene grindstone/mortar technology used to process a wide range of resources and not simply seeds. Likewise, several researchers have argued that morphological variation may more simply relate to the availability of raw materials, the degree of curation or the state of preservation, rather than intentional design relating to production efficiency (Balme et al. 2001; Gorecki et al. 1997; Smith 2015b; Veth and O’Connor 1996). In their view, seedgrinding may not be a late Holocene development at all, but part of an economy that

49 stretches well back into the Pleistocene in some areas (Balme et al. 2001; Fullagar and Field 1997; Hayes 2015).

Of the matters which can be informed by grindstone technology, two issues remain central to the debate— those of form and efficiency. For example, if it can be confirmed that the morphological variation in some grindstone types may be ascribed to accumulated use wear over time rather than design factors intended to improve efficiency, then the dichotomous distinction between formal and amorphous grindstones needs to be questioned for its usefulness (Balme et al. 2001:5; Veth and O’Connor 1996:20). A flat slab classified as an amorphous grindstone could, with use, develop a groove or dish and thus be reclassified as a formal implement. If a large grooved millstone confers a significant productive advantage over other grindstones then it may be a specialised implement and its introduction into the technological toolkit may indicate a recently developed reliance on a seed based diet. However, if it only confers an incremental advantage, it may instead be merely a minor improvement to existing old and widely used practices. If grindstone types other than the specialised millstone are able to efficiently process seeds, then arguments that seeds may have been a more widespread component of diets than presently believed are supported (Balme et al. 2001; Fullagar and Field 1997; Hayes 2015).

These and other elements of the debate can be informed by carefully designed experiments producing controlled quantitative data.

Methods and Research Design A central assumption underlying much of the debate outlined above is that the millstones as described by Smith for Central Australia (1985) were more efficient than other types of grindstones for processing seeds (Smith 2015b), and that it is for this reason that a particular grindstone morphology emerged in response to intensified use of seeds in the Late Holocene. This assumption could be tested using either ethnographic or experimental data. However, while the ethnographic literature documents several timed grinding sessions which could in theory be used to determine the efficiency of different types of grindstones, much of this data is qualitative and highly variable in quality and nature (Brokensha 1975; Cane 1984; O’Connell and Hawkes 1981; O’Connell et al. 1983; Smith 1985; 1986; 1989). For this reason we have opted in this paper for an experimental approach to determine the differential efficiency of grindstones in terms of human labour and quantity of seed processed in a given time with each type of grindstone. To this end, we performed milling experiments to determine whether basal grindstones of varying size and configuration perform at different levels of efficiency when processing seeds. In investigating the millstones, some preliminary consideration was also given to such matters as the effects of the muller or top stone, the seed being ground and whether the grinding action used was reciprocal or rotary. In general, researchers using grinding technology to investigate high- level matters such as efficiency, diet breadth, nutrition and so on have not isolated the effects of these variables, all of which could influence outcomes. Our experiments aimed to generate comparative data between the grindstones. While the raw data are valuable, outcomes can only be taken as indicators of the possible performance of similar grindstones in a traditional setting.

Grinding

The primary purpose of the experiments was to establish the efficiency of a number of replicated basal grindstones by measuring their output of ground grain under a variety of circumstances.

50 Ten basal millstones ranging from very small (20 cm Over-all Length- OAL) to large in size (57 cm OAL) and with either different functional surfaces or degrees of simulated wear were replicated, plus a mortar. The dimensions of the experimental grindstones are shown in Table 1 and the grindstones are photographed in Figure 2A and B. They consisted of unused and used (artificially aged) pairs of large open groove, large dish, medium trough, small open groove and very small trough ‘millstones’ plus the ‘shallow cup’ mortar. The grindstones were numbered from 1 to 11 and are referred to by number and abbreviated description, e.g. ‘1-LGU’ (Large, Groove Used).

An open groove completely traverses the block on which it is made whilst a trough is closed at both ends. A dish is a larger, basin like depression usually encompassing much of the surface. The definitions adopted here are far from universally accepted with numerous alternatives in use. An estimate of the degree of simulated wear created is shown in the Wear column of the table.

Each created grinding surface has been regarded as a separate grindstone despite some being manufactured on the same block of stone. For convenience, we have adopted the terms ‘New’ to represent, at commencement of the experiment, an implement with a natural rough surface and ‘Used’ when the experimental piece was given a manufactured groove. The terms may, in most contexts, be regarded as analogous to ‘formal’ and ‘amorphous’.

The grindstones were reproduced in large blocks of Helidon sandstone from Southeast Queensland. This is a resilient and durable quartzose sandstone with sharp, medium-fine, tightly interlocking, well cemented grains widely utilised in construction (including the original buildings of the University of Queensland campus). For a detailed lithological analysis, see Richards 1911:201-2. Figure 1 shows grain structure.

Figure 1. Image of natural raw surface of Helidon sandstone at 20X magnification. Depth-of-field was enhanced using Zarene Stacker software.

A 230mm industrial angle grinder and carborundum pads were used to form the depressions. As they are constantly evolved by use, grindstones have a virtually infinite range. Some 170 descriptions of grindstones in the literature were examined as models and the selection replicated is considered to be representative of the majority of commonly reported sizes and types of millstone. However, it is stressed that this is a preliminary assessment and the selection is far from exhaustive, especially if areas other than the dry interior are considered.

51 Grindstone sizes are commonly described in relation to the overall length of the slabs on which they are developed. However, it is the grinding surfaces which are critical. The largest slab was 57 cm long but the longest depression utilised was 43 cm. The smallest slab was 20 cm long and the shortest depression, excluding the mortar, was 16 cm. Ethnographic examples outside this range are known but are not common. A depression of 43 cm was arrived at as an informal biometric limit as it represented a comfortable extension of the arms for the experimenters- both males of above average height.

Block Depression

Number Grindstone Description Wear Lengthcm Widthcm Depthcm Weightkg Type Lengthcm Widthcm Depthcm Area cm Comments 1 Millstone Large, Groove Used LGU 25% 47.0 46.0 4.0 17 Groove 43.0 13.0 0.1 582 Large slab shared with Nos.2 and 6. 2 Millstone Large, Groove New LGN 0% 47.0 46.0 4.0 17 Groove 43.0` 13.0 0.0 559 Large slab shared with Nos.1 and 6. 3 Millstone Small, Groove Used SGU 15% 25.0 25.0 11.0 16 Groove 23.0 10.0 0.7 235 Block shared with Nos.4 and 5. 4 Millstone Small, Groove New SGN 0% 25.0 25.0 11.0 16 Groove 23.0 10.0 0.0 230 Block shared with Nos.3 and 5. 5 Mortar Block MtU 10% 25.0 25.0 11.0 16 Cup Dia. 13.5 2.1 150 Block shared with Nos.3 and 4. 6 Millstone Large, Dish New LDN 0% 47.0 46.0 4.0 17 Basin 43.0 29.0 0.0 1066 Large slab shared with Nos.1 and 2. 7 Millstone Large, Dish Used LDU 4% 57.0 40.0 4.0 18 Basin 43.0 29.0 2.0 1085 Large slab shared with Nos.10 and 11. 8 Grindstone Very Small, Trough New VTN 0% 20.0 13.0 4.0 3 Trough 16.0 9.5 0.0 92 Small slab shared with No. 9 9 Grindstone Very Small, Trough Used VTU 25% 20.0 13.0 4.0 3 Trough 16.0 9.5 1.0 95 Small slab shared with No. 8 10 Millstone Medium, Trough New MTN 0% 57.0 40.0 4.0 18 Trough 34.0 14.0 0.0 374 Large slab shared with Nos.7 and 11. 11 Millstone Medium, Trough Used MTU 55% 57.0 40.0 4.0 18 Trough 34.0 14.0 2.1 390 Large slab shared with Nos.7 and 10. Table 1 - Specifications of Grindstones - all made from medium-fine grained Helidon, Queensland sandstone.

The role of mortars in seed processing is not entirely clear. Some grains were crushed into a coarse meal on mortars but with other seeds, this was only an initial processing step before further grinding on a millstone (Smith 1985:27). McCarthy is clear that in areas without millstones, mortars ‘were used for pounding seeds and nuts, and grinding them into the flour which was mixed into a paste and baked into a damper loaf’ (1976:63). Accordingly, the mortar has been evaluated as a small seed-grinder but its important additional pounding functions should be kept in mind. As with the replicated example, many Australian mortars have a relatively shallow depression (Smith 1985:27) compared to some international ‘deep’ mortars in which the depth of the cup is equal to, or even greater than, its diameter. In Australia, the distinction between mortars and some other implements, especially the smaller basin millstones, may not always be clear (Kraybill 1977:505; Smith 1985:29).

A. B Figure 2. Replicated grindstones. A. Nos. 1, 2, 4, 5, 7 and 9. B. Nos. 3,6,8,10 and 11. Chalk lines show approximate functional areas and highlight the shapes of depressions. (Photographs by John Mildwaters).

The literature regularly refers to seeds being prepared by a wet grinding process but Aboriginal millers also ground grain to produce a dry powder product (e.g. Gregory 1887:132; Newland 1920-21:14; Smith 2015b). Accordingly, the experiment emphasised wet grinding, 187 sessions, but included around a quarter of the total (66) of dry grinding sessions for completeness and comparison. Adding water will affect weight of output. As a check on the technical procedures adopted, a comparison was made between the input of Grain Used and the outputs of Wet Meal Produced and Dried Meal Produced. The results of Pearson’s r test for the Grain Used / Wet Meal relationship was 0.915 and for Grain Used / Dried Meal was 0.939 (both significant at the p = 0.01 level, 2-tailed). These high correlations suggest that both Wet Meal and Dried Meal have a similar strong relationship to Grain Used and that adding water does not affect the reliability of results.

Each grinding session comprised of 8-12 loads of grain on the grindstone and lasted for 10 minutes. A total of 280 grinding sessions were performed, representing over 46 hours of actual grinding. After excluding the sessions involving native seeds, tests and cleaning the data, 253 sessions of ten minutes each were available for analysis; 23 sessions or 3.8 hours of grinding for each grindstone. Variations within the 23 grinding sessions were kept constant for each grindstone. For example, each grindstone had 17 wet grinding sessions and six dry.

The selection of the duration of 10 minutes for a grinding session was a largely arbitrary decision. However, it was considered short enough to enable a session to be readily excluded if a mistake, such as spilling grain, adding too much water etc. was made. In addition, the 10 minute session with an enforced respite for measuring and weighing at its

54 end, physically allowed the researchers to continue milling for lengthy periods without requiring additional rest.

As extrapolating from the outputs of 10 minute sessions to a more useful measure of ‘hours per kilogram’ risked amplifying small relative differences in individual outputs, 500 grams of millet was more or less continuously ground as a check using millstone No. 10 (Medium, Troughed, New). Actual grinding time was 79 minutes but rest breaks were necessary as grinding is hard work. Grinding time for each 100 grams of grain varied considerably from 10.2 minutes to 23.0 minutes. The shortest time was a result of intentionally grinding as fast as possible. However, this led to exhaustion and eventually, as short breaks did not overcome tiredness, to the longest time of 23.0 minutes. On average, the intensity of the grinding was intended to match the brisk, but unhurried, rhythm of the experimental grinding. Because of the confounding effects of tiredness, there is little point in attempting to continuously grind more than, say, 200 grams.

The 500 grams was converted to wet meal at a rate of 1.64 hours per kilogram. In the 10 minute experimental sessions, 723 grams of millet were converted to wet meal with this grindstone at a very similar rate of 1.80 hours per kilogram. Accordingly, the 10 minute sessions would not appear to adversely influence results.

A variety of mullers or topstones were used consisting of flat sandstone blocks or water-worn pebbles. Generally, the mullers were not precisely matched to, or paired with, the millstones although obvious incompatibilities between the concave groove profiles of millstones and the convex profiles of the grinding surfaces of the mullers were avoided. Where a muller was judged incompatible, no further grindings with that particular millstone-muller combination were attempted.

A supply of grain was weighed on electronic scales before and after each grinding session to establish the weight of grain processed — that is, the grain ‘Used’ amount.

At the completion of each 10 minute grinding session, the product, either dry meal (‘Flour’) or ‘Wet Meal’, was also weighed. Before weighing, the wet meal was worked between the hands into a ball or cake. The dry meal or flour was later mixed with water, also formed into cakes and weighed. Whilst able to be treated separately, the dry meal could thus be combined with the wet meal and also treated as wet meal (see above). The cakes were allowed to dry and then weighed, giving the weight of ‘Dried Meal’.

It was found that, for an experiment of this magnitude, there was no precise way to determine exactly when the grain was completely ground, especially as grinding characteristics varied between seeds. No measures of fineness of grind could be located in the ethnographic record. However, as ‘cakes’ are frequently mentioned, it was considered that ‘fineness’ which would allow the meal to adhere to itself and form cakes would be a worthwhile proxy for the fineness of the ground meal. Screening of the product was considered but would have extended the experiments beyond available resources. With the dry grindings, adherence could not be determined until after the fact. In practice, a definite ‘feel’ for the quality of the grind, within a limited range, developed with experience and only a very small number of sessions resulted in meal that would not form into cakes. Whilst output was the primary consideration, it would have been misleading not to take some account of fineness of the grind as clearly, less time is required to produce a rough meal than a fine flour. Accordingly, a record of subjective fineness was created. A rank scale of 1 to 5 was devised to capture these differences in texture and adherence (see Table 2). The scale is ordinal in that meal recorded as a ‘3’ is more finely ground than that recorded as a’2’ but meal shown as, for example, ‘2.5’ only demonstrates that an equal number of ten minute sessions recorded as twos and threes formed the mean.

55 Rank Key Description 1 Sandy Meal will not adhere and form a cake. Proportion of unground or partially ground grains too high. Requires regrind.

2 Gritty Difficult to form meal into cakes which tend to crack and 'flake' apart. Cake will not 'sop up' loose meal and merge it into the cake.

3 Soapy Acceptable. Forms into a cake with some care for cracks and incorporates most loose meal.

4 Plastic Much like play dough. Readily forms into a cake and incorporates almost all loose meal but with some tendency to adhere to hands rather than to itself.

5 Sticky Much like sticky, wet clay or putty. Adheres to itself more than hands and 'curls' on the muller. Forms cake without need for 'rolling' between palms. Effectively sops up all loose meal. Table 2 - Classification of subjective degrees of 'fineness of grind'.

Waste

During the experiments it became obvious that wastage may have been an important factor in assessing the efficiency of some grindstones. Waste can consist of any combination of whole dry grains (which may, for example, bounce off the surface as the grindstone is loaded), partially ground wet or dry grains through to wet or dry meal of varying degrees of fineness and stickiness. Whilst such matters as transfers between containers to record experimental results may well have accentuated waste, it was considered prudent to specifically account for it. Accordingly, all waste was collected from the implements and a groundsheet and weighed immediately after grinding.

Seeds

Native seeds proved difficult to obtain in sufficient reliable quantities to conduct a large experiment. Accordingly, commercial domesticated seeds were utilised as proxies. After investigation, the following three grains were adopted. Gatton green panic (Panicum maximum) is a small, soft and light (370 grams per litre approximately) grass seed with similar morphology and texture, and related at genus level, to the widely utilised native millet (Panicum decompositum). Panic is soft and spongy and its grains do not shatter. It grinds into a light, voluminous fibrous mass with limited adherence unless very finely ground. As can be seen from the photograph (Figure 3A), the native and commercial seeds are very similar. Sufficient native millet could not be obtained for comparison but it is reasonable to assume that grinding characteristics are similar and the commercial panic should be a satisfactory small seed substitute for the experiments.

56 A B

C Figure 3 - A: Native millet (L) and Gatton panic (R) seeds showing similar morphology. B: White french millet seeds. C: Seeds of mulga (L), Queensland red sorghum (C) and elegant wattle (R) showing similar grain size. Background grid is 2mm. (Photographs by John Mildwaters).

White French millet is a hard, medium sized domesticated seed weighing approximately 620 grams per litre (Figure 3B). Cleland identified over a dozen food grasses with seeds ‘about the size of millet’ (1966:146; also Isaacs 1987) and O’Connell et al. confirmed that the millets are ‘morphologically and taxonomically similar to those domesticated’ (1983:99). White French millet should represent a viable proxy for medium sized native seeds including the larger grass seeds. The largest, arm grass millet (Brachiaria miliiformis), (O’Connell et al. 1983) is more elliptical in shape (about 2.5 by 1.5 mm) (Brook Clinton & Maggie Nightingale, Australian National Herbarium, Pers. Comm.) but of similar volume. The smaller acacias such as raspberry jam tree (Acacia acuminata) and dead finish (A. tetragonophylla) are also of comparable size.

Sorghum is a relatively dense domesticated grain which weighs approximately 850 grams per litre. The variety selected, Queensland red sorghum (Sorghum bicolor), is closely related to the native sorghums S. laxiflorum and S. macrospermum (Dillon et al. 2004). However, only one relatively late reference to these grains being ground for food has been located (Arndt 1961:109), hence it was deemed unnecessary to establish a proxy for the native sorghums. Regardless, red sorghum is a hard seed of similar size to Acacias such as mulga (Acacia aneura) and elegant wattle (Acacia victoriae) (Figure 3C) although it bears no genus level association with the acacias. These two widely used Acacias have respectively been described as being both soft seeded (Horne & Aiston 1924; Cleland & Tindale 1959) and

57 hard seeded (Smith 1985). Subjectively, the native seeds appear morphologically similar to the sorghum, thus the commercial sorghum would appear a reasonable substitute for experimental use.

Small amounts (140g) of mulga and elegant wattle seeds were obtained and were ground for comparison. Grinding the mulga revealed the sample obtained to be a hard seed, but non- brisant. It was near impossible to grind (resembling a hard rubber pellet under an even harder coating) with grains deforming but not shattering under the muller and depositing a heavy waxy reflective polish on both muller and millstone. A further sample of the mulga was then parched in an oven at 200°C for 15 minutes. The seed became soft and brisant and slightly easier to grind than sorghum. Sorghum is thus an acceptable proxy for parched, but not raw, mulga.

In view of the experience with mulga, the grindstones were regularly inspected and cleaned when necessary to ensure that an accumulation of polish did not affect abrasiveness.

According to O’Connell et al. (1983:92), hard Acacia victoriae seed must be cracked before it can be ground. However, it was found that raw seed could be ground, with some difficulty, and without depositing a waxy gloss. It formed a sticky but relatively coarse meal. After parching, its grinding characteristics improved, though it was still more difficult to grind than raw sorghum, but sufficiently similar to regard sorghum as an adequate proxy for parched A. victoriae.

While some 43 different acacias are recorded in the literature examined as being ground for food, we consider that sorghum provides a satisfactory proxy for generic large, hard seeds as utilised in traditional economies. However, in the absence of at least genus level matching, fewer grindings using sorghum were performed than with the two seeds closely related to native grasses.

Each grindstone was used for 23 sessions. Millet was ground in 11 sessions, panic in seven and sorghum in five- a total of 253 sessions which were available for analysis. A further 27 sessions were excluded for various reasons including those which were conducted with native grains. The weights of each grain ground on each grindstone are shown in Table 3.

Grain Processed by Weight- All Grindstones

Weight (g) Grindstone Millet Panic Sorghum 8-VTN 442 97 192 731 9-VTU 438 132 189 759 5-MtU 652 153 229 1034 3-SGU 648 144 245 1037 4-SGN 707 166 267 1140 10-MTN 723 197 281 1201 11-MTU 780 212 328 1320 6-LDN 1162 279 365 1806 2-LGN 1189 226 419 1834 1-LGU 1264 263 382 1909 7-LDU 1444 306 429 2179 Total 9449 2175 3326 14950 Table 3 - Weight of each grain processed on each grindstone.

58 Results Full details of the experimental grinding results for each grindstone are shown in Table 4 and Figure 4. Mean amount of grain used in grams is the best currency for processing efficiency of each grindstone given all experiments were 10 mins in duration. The grain used should, of course, be considered in conjunction with the amount of grain wasted as there is a relationship, if not a direct one, between the two. However, as there is no ethnographic clarification of waste treatment and the relationship is indirect, Waste has not been deducted from Grain Used in calculations. As is common in the ethnographic literature, Wet Meal has been used for calculating labour hours.

Performance 160 145 140 123 122 117 120

100 95 89 86 82 80 83 79 79 80 65 67 Grams 63 64 59 59 57 59 60 52 47 50 42 45 45 45 46 46 40 32 33 29 29 32 22 25 21 20 13 13 13 10 8 10 10 11 8 7 11 0

Used Wet Meal Dried Meal Waste

Figure 4 - Grain Used, Wet and Dry Meal Produced & Waste (g) of the replicated grindstones (arranged smallest to largest, left to right).

59 H/Kg N Used Flour Wet Dried Wet Waste Waste% Fineness Total 253 59 37 86 45 2.9h 11 19% 3.7 8-VTN 23 32 28 42 22 5.1h 13 42% 3.1 9-VTU 23 33 22 47 25 4.2h 13 40% 3.2 5-MtU 23 45 28 59 29 3.7h 10 22% 2.3 4-SGN 23 50 23 59 29 3.7h 21 41% 3.5 3-SGU 23 45 26 63 31 3.7h 13 28% 3.3 10-MTN 23 52 37 82 41 2.5h 8 15% 4.4 11-MTU 23 57 38 89 44 2.3h 10 18% 4.1 2-LGN 23 80 49 117 65 1.7h 10 13% 4.3 1-LGU 23 83 53 123 68 1.6h 11 13% 4.2 6-LDN 23 79 53 122 63 1.7h 8 11% 4.3 7-LDU 23 95 57 145 77 1.4h 7 8% 4.4 Table 4 - Overall mean performance of all grindstones (in increasing functional surface area).

The first eight grindstones discussed are of either ‘trough’ or ‘groove’ type worked with a reciprocal motion. Classification of grindstones is uncertain and it may be that a distinction between the two types is unnecessary. However, until the matter is further clarified, we consider it prudent to maintain the distinction.

Grindstones Nos. 8 (Very-small Troughed New) and 9 (Very-small Troughed Used) have functional grinding surfaces of only 92 and 95 cm² respectively. Mean grain Used each session was 32g (N = 23) and 33g (N = 23) respectively but they each wasted 13g or over 40% of the grain processed. Both produced meal of below average fineness but the fineness was sufficient for the meal to adhere and form into cakes. Producing a kilogram of Wet Meal required 5.1 and 4.2 hours of grinding respectively.

Millstones Nos. 4 (Small Grooved New) and 3 (Small Grooved Used) have functional areas of 230 and 235 cms² respectively. Performance was comparable. Grain Used was 50g (N = 23) and 45g (N = 23) but the new stone wasted more grain— 21g or 41% as against 13g or 28%. The millstones were each able to produce a kilogram of Wet Meal in 3.7 hours. Cakes formed readily. Opinions vary as to whether these small grindstones should be regarded as formal millstones but, in view their performance, we consider that classification reasonable.

The larger troughed grindstones Nos. 10 (Medium Troughed New) and 11 (Medium Troughed Used) have depression lengths of 340mm and functional surface areas of 374 and 390 cm² respectively. They both would be regarded as formal implements once the incipient groove on the ‘new’ grindstone became apparent. Grain Used at 52g (N = 23) and 57g (N = 23) was moderately higher than the smaller grindstones but average wastage was lower at 9g or 17%, fineness of grind was high and only 2.5 and 2.3 hours were needed to produce a kilogram of Wet Meal.

Performance of the last pair of ‘grooved’ mills, Nos. 2 (Large Grooved New) and 1 (Large Grooved Used) with open groove lengths of 430mm and functional surface areas of 559 and 582 cms² was similar between the pair. Grain Used was 80g (N = 23) and 83g (N=23), waste was no more than 11g or 13% and fineness of grind was high. These ‘specialised’ millstones were able to produce a kilogram of Wet Meal in only 1.7 and 1.6 hours.

The mortar (No. 5) has a different configuration to a grindstone and its cup depression was created in the top surface of a sandstone block. It would be regarded as a formal implement. The depression has an approximate functional surface area of 150 cm². It processed a

60 reasonable 45g of grain each session (N = 23). Waste was a creditable 10g or 22 per cent. Fineness of grind at 2.3 was the lowest of the grindstones. The mortar was able to produce a kilogram of meal in 3.7 hours.

The final pair of millstones, Nos. 6 (Large Dished New) and 7 (Large Dished Used), are of a dished configuration and are operated with a rotary or circular, rather than a reciprocal or to- and-fro, action. Like the large grooved millstones, these implements, once wear was apparent, would be classified as formal implements.

In length, they are similar to the large grooved pair but the functional surface areas are much greater—1066 and 1085 cms² respectively. Grain throughput was 79g (N = 23) and 95g (N = 23) respectively (mean 87g). Wastage averaged an excellent 8g or 11% and 7g or 8% and fineness of grind was high. Time to produce a kilogram of Wet Meal was 1.7 hours and 1.4 hours, the lowest of all grindstones.

Labour hours to produce one kilogram of wet meal for all implements are shown in Figure 5 and Table 4.

Grinding Hours per Kilogram Wet Meal

Mean 7-LDU 6- LDN 1- LGU 2- LGN 11-MTU 10- MTN 3-SGU 4-SGN 5.MtU 9- VTU 8- VTN

0.0 1.0 2.0 3.0 4.0 5.0 6.0

H/Kg

Figure 5 - Grinding hours for each grindstone to produce a kilogram of Wet Meal.

Discussion The individual grindstones performed at varying levels of efficiency.

Very-small Trough New and Used

The tiny grindstones, Nos. 8 and 9, performed similarly. At only 16 cm in length, they are very unstable whilst grinding and using two hands to grind for any length of time is difficult. They are also particularly sensitive to the selection of a compatible muller. Because of the

61 limited working area available, the muller needs to be used almost as a pestle with an action that could be said to combine crushing and grinding. Large mullers tended to glide or roll over the grains—a bed-of-nails effect— whilst small mullers required considerably increased downward force and sometimes risked finger injury.

Although these very small grindstones are clearly capable of grinding seeds, with their low output and high wastage, they suggest that morphology does impose some practical lower limits on the size of a grindstone used for the regular milling of grain (whilst not necessarily precluding emergency intermittent use or use for preparing ochre, medicines and so on). It is difficult to envision these very small grindstones being used for long-term heavy-duty processing of seed staples as their times to produce a kilogram of wet meal are towards the longer times recorded in the literature.

Mortar

Again it is difficult to picture the mortar as a primary seed processing implement. Whilst, on average, fineness of grind was adequate, the mean concealed a number of sessions which required a regrind to achieve adherence. Of course, in traditional usage, the need for an occasional regrind may well have been only a minor inconvenience. In addition, when wet meal production is considered, the mortar is capable of producing a kilogram of wet meal in 3.7 hours, well within the range of the ethnographic grinding observations. The mortar also has a desirable pounding capability and this widely distributed artefact was clearly a flexible and valuable multi-use implement with secondary seed processing capability. As such, the potential for seed-use exists in areas where the mortar was part of the common toolbox.

Smith mentions that the depression of a mortar may be initiated by pecking (1985:28), a clear indication of intentional design.

Small Grooved New and Used

The two small millstones, numbers 3 and 4, at 25 cm in length, are at the lower end of what has sometimes been regarded as specialised millstones (Smith 1985). Their performance was similar to the mortar but they were generally able to grind a finer meal.

The waste performance of the SGN (No.4) stone is anomalous. It was formed from the shallow side of the block containing the mortar and was without any coping surrounding the grinding surface. Coping can be effective as a reservoir for unground grain and a collector of seeds which would otherwise escape sufficiently to be lost. For the sake of consistency in the experiments, no conscious advantage was taken of the surrounds of any stones but, in the normal rhythm of grinding, it is inevitable that a portion of potential waste will be retrieved from the coping. The difference in waste performance can probably be attributed to this factor. Smith (1985) suggests that millstones of less than 300mm in length may not be efficient. However, if formed in a stable slab with adequate coping, it is considered that the 250mm grindstones meet the requirements of a specialised millstone.

Medium Trough New and Used

The larger troughed millstones, Nos. 10 and 11, slightly longer than Smith’s minimum with groove lengths of 34 cm, appear to provide only a moderate increase in performance over the smaller grindstones. However, this apparent slight underperformance may be misleading as, when wastage and fineness of grind are taken into account, these millstones comfortably out-performed the smaller implements. This justification does not necessarily remain valid when the comparison is made with larger millstones.

62 Large Grooved New and Used

The two large grooved millstones (1 and 2) , with overall lengths of 47 cm and groove length of 43 cm, are considered representative of the larger millstones that could still be comfortably worked by an average sized woman. They were both able to produce a kilogram of meal much faster than the times used in most theoretical calculations. Davidson and McCarthy considered them specialised formal implements (1957:438-9) and, as expected, their overall performance was considerably higher than the smaller grindstones.

Large Dish New and Used

Dished millstones are not well documented in the ethnographic literature although McCarthy states that they are ‘very widely distributed throughout the [more south-eastern arid and semi-arid desert areas of the] interior, and occasionally examples are found along the east coast’ (1976:59). The production results of the dished millstones (6 and 7) are slightly unexpected in that they marginally outperformed the large grooved mills and are thus the most productive of the millstones. However, the effective grinding area of the dished mills is much larger than that of the grooved mills and to achieve a nominal increase in throughput and reduction in wastage, almost twice the raw material was required to manufacture the millstones. The slight increase in production would likely not warrant the increased raw material cost in most situations. These dished millstones are very forgiving of the muller used and perform well with a wide variety of topstones.

Size

The factor that had the largest bearing on results was grindstone size and performance improved as the size of the grindstones increased. The large grindstones clearly outperformed the small units in all measures of efficiency. The relationship between size and efficiency is linear but a critical threshold could not be precisely discerned.

Sessions with the five smaller grindstones, on average, Used 41g (N = 115), Wasted 14g or 34%, ground to a fineness of 3.1 and required 4.1 hours to produce a kilogram of Wet Meal. The six medium to large millstones Used 74g (N = 138), Wasted only 9g or 12%, ground very finely at 4.3 and produced a kilogram of Wet meal in only 1.8 hours (Table 5). The difference in yield between small and large grindstones is also significant (t (253) = -8.589, p ≤ 0.01). Thus size is paramount.

aste % aste

N Used Flour Meal Wet Dried Meal Hrs/Kg Wet Waste W Fineness Small-3,4,5,8,9 115 41 25 54 27 4.1h 14 34% 3.1 Large- 1,2,6,7,10,11 138 74 48 113 60 1.8h 9 12% 4.3

Table 5 - Performance of small and large grindstones.

Overall, the experiments support Mauldin’s contention that grinding efficiency is positively related to grinding surface (1993:319). Functional surface area is critical and there is a relatively direct relationship between it and productivity. The relationship between production and functional surface area is illustrated in Figure 6. The mortar and the dished millstones

63 have been included and, despite their different configurations and grinding actions, do not materially distort the relationship.

Grain Used and Functional Surface Area 1200 120 1066 1085

1000 95 100 83 80 79 800 80

57 582 600 559 60 50 52 45 45 374 390

400 32 33 40 GrainUsed grams 235

FunctionalArea Sqare Cms 230 200 150 20 92 95

0 0

Surface Used

Figure 6 - Functional Surface Area and Grain Used.

New vs Used

In an endeavour to verify the postulated superiority of unused grindstones (Gorecki et al. 1997; Veth and O’Connor 1996), the grindstones were divided into two groups and the performance of the ‘new’ replicated grindstones was compared to the ‘used’ (see Table 6). Rather, our experiments tend to give some slight support to Warner who reported that millers in a non-grass seed area preferred a grindstone with an established depression. Whilst the real differences are clearly not large, the used grindstones did slightly out-perform the new in all the meaningful measures of performance. However, this is not to suggest that the groove in itself has an element of design functionality or that the groove in similar traditional sandstone millstones is likely to be anything beyond a product of use-wear. We were unable to establish that the difference was statistically significant, [for Grain Used, t (115) = -.905, p = 0.367)].

N Used Flour Meal Wet Meal Dried Wet Hrs/Kg Waste % Waste Fineness New 115 58 38 85 44 2.9h 12 21% 3.9 2.4.6.8.10 Used 115 63 39 93 49 2.6h 11 17% 3.8 1,3,7,9.11 Table 6 - Performance of new and used grindstones.

It should be understood that if well-worn grindstones do have a slight advantage, this is only likely to apply where a muller fully compatible with the groove profile of the grindstone is available and the groove is not deep enough to cause interference with the grinder’s fingers. Use of an ill-fitting muller means that grain is only partially ground or remains unground. With millstone No. 11, which was estimated to be around 55 per cent worn, it was noted that, whilst performance was not adversely affected, partially ground grain tended to accumulate at the ends of the groove and the muller sometimes snagged and twisted which destroyed the rhythm of that particular grinding stroke and occasionally injured the miller’s fingers. This millstone still had many hours of productive use ahead but it was quite sensitive to the selection of topstone and needed to be operated with some additional care.

Seed Type

Along with grindstone size, the type of seed being processed has a major effect on productivity. The seed being processed affected performance of all millstones- see Table 7. There was a statistically significant difference between groups as determined by One-Way ANOVA [F (2,250) = 76.668, p ≤0.05)]. A Tukey post-hoc test revealed that a statistically significant difference applied to comparisons between all three seeds—millet 78.1 ± 34.4g, p ≤0.01; panic 28.3 ± 13.9g, p ≤0.01 and sorghum 60.5 ± 25.2g, p ≤0.01.

The small and soft, but spongy, panic was the least efficient to process. Only 28 grams were able to be ground per session—roughly half that of sorghum (60g) and little more than a third that of millet (78g). Whilst the millet and sorghum produced a kilogram of Wet Meal in less than two and a half hours, the panic required over four hours. If the proxy grains are indicative, having to process a particular seed could well virtually double a person’s daily labour and thus influence economic decisions as to whether seeds should be included or excluded from a diet.

The few additional grindings able to be performed with native seeds, together with the variations in grinding times recorded by Cane (1984) when using a variety of seeds, suggests that further planned work with traditional seeds is essential. The proxy seeds can establish valid comparisons between grindstones but are no substitute for establishing the particular grinding characteristics of the native seeds.

Grain N Used Flour Wet Dried Hrs/Kg Waste Waste Fineness Meal Meal Wet % Panic 77 28 16 61 26 4.3h 7 26% 3.7

Millet 121 78 57 102 55 2.2h 15 19% 3.5

Sorghum 55 60 39 87 49 2.4h 9 15% 4.3

Table 7 - Mean performance of all grindstones with various grains.

Grinding Action

In comparing rotary and reciprocal grinding actions, Adams has suggested that the rotary action may be more stressful on the shoulders and elbows than a reciprocal action (1994:79) and subjectively this is supported as there is some difficulty in maintaining muller pressure through the complete grinding cycle. However, our experiments confirm the dished implements are less tiring due to a greater range of allowable motor actions (Adams 1994:79) and easier to use as there is little risk of lost grain and the circular motion naturally repositions unground grains in proximity to the muller. Occasionally reversing the direction of rotation from anti-clockwise (for a right handed miller) to clockwise further helps sweep grain

65 back to the muller without losing grinding rhythm. These factors should confer some advantages in efficiency.

For these reasons, it was expected that the dished millstone/ rotary action combination would be more efficient than a grooved millstone/ reciprocal action. This intuitive expectation was only partially supported by results- see comments on the individual grindstones. For a comparable length, the rotary action encompasses a wider area and thus dished millstones have a larger functional grinding area for a comparable length. Meaningful comparisons are difficult and the relative efficiency of reciprocal versus rotary grinding was unable to be resolved with the present experiments due to the confounding influence of the much greater grinding surface area of dished millstones.

We consider this a worthwhile issue to pursue as dished millstones were common in the seed-using eastern arid grassland areas. Replicated dished millstones with grinding surface areas comparable to grooved millstones would be required.

Wet or Dry Grinding

As mentioned earlier, the conventional interpretation is that wet grinding was the usual, and critical, technique for a number of reasons including that the water collected the fine flour and prevented loss by wind (M.A. Smith, Pers. Comm.) and that wet grinding increased digestibility (Harney 1951). However, in historical and ethnographic accounts, there are numerous references to dry grinding (e.g. Roth 1897; Newland 1920-21; Horne & Aiston 1924; Cleland & Johnston 1933). Our investigation suggests that a further advantage of wet grinding is that more seed can be ground in a session. With every grindstone, the weight of grain ground wet exceeded the amount ground dry. Generally, only about 60 per cent of the wet weight was able to be ground dry (Table 8). These results support the impression that, when possible, the wet grinding method would likely have been preferred.

Wet & Dry Grinding- Grain Used % Dry to Wet Dry Wet 8-VTN 33 29 88 9-VTU 37 22 59 5-MtU 50 31 62 3-SGU 51 29 57 4-Sgn 56 32 57 10-MTN 57 38 67 11-MTU 64 38 59 6-LDN 85 61 72 2-LGN 90 52 58 1-JGU 93 56 60 7-LDU 107 60 56 Mean 66 41 62 Table 8 - Weight of grain used (g), wet and dry grinding.

General

The development of a depression or groove has implications for the formal/ amorphous question. At the conclusion of the experiments, all the ‘new’ replicated grindstones had developed at least limited depressions. For Nos. 4, 6 and 10, after 3.8 hours of grinding, these were incipient only and to a maximum depth of 1.5 mm. No.2 was used for 5.2 hours

66 and developed a distinct groove to a depth of 3.0 mm. The anomaly was the Very Small Troughed grindstone No. 8 which, after 3.8 hours of grinding, had developed a depression to a maximum depth of 5.5 mm. An obvious explanation is that the small slab on which the grindstone was made is softer and not representative of the sandstone used for the other implements. However, grindstone No.9 was made on the obverse face of the slab and only deepened by 2.0 mm with the same hours of grinding so this seems unlikely. The data are too preliminary to offer firm conclusions but it is possible that the evolution and longevity of a grindstone may be determined by factors other than the raw material from which it was made and the time it was used. If so, possible differential wear regimes call into further question dichotomous distinctions between formal and amorphous grindstones.

The small grindstones tested all displayed some instability whilst grinding due to lack of inertia, especially with the large, hard grained sorghum which required greater crushing pressure to fracture the grains. Whilst methods to stabilise grindstones were practiced (e.g. Cane 1989:106; Horne and Aiston 1924:55), it is considered that instability may become a major problem as size and weight are reduced. The two very small grindstones provide insight into the likely performance of millstones such as those discussed by Veth and O’Connor (1996:20) and the ‘small personal’ type of grindstone described by Mulvaney (1998:88-9).

The focus of the experiments was on the performance of the bottom stones. However, more attention could have been devoted to the topstones. Whilst the muller used in each grinding case was recorded, individual mullers were not restricted to particular grindstones. It would have been preferable for future examination of the performance of the topstones if each muller had been used with only one grindstone so that, for example, any differences between sandstone blocks (which seem to ‘shear’ grains) and water worn pebbles (which appear rather to ‘roll’ grains) would have been apparent and the development of incipient wear patterns would have been preserved. Regardless, a separate full investigation of top stones is required.

It is difficult to assess the proficiency of the experimenters as millers. Whilst it is most unlikely we achieved the skill of a person with a lifetime of experience, we worked in a brisk but unhurried manner in a workshop with the grindstones on a large table covered by a plastic groundsheet and without interruption during the 10 minute grinding sessions. It is improbable that these ideal conditions would have been enjoyed in practice, with other duties and distractions intruding on grinding time. In addition, grinding is a messy activity and ground wet meal adheres to every surface—hands, muller and grindstone—and is easily thrown off the grindstone and lost if not saved by the groundsheet. It may be that losses in the field were much higher than we experienced. Of note is that in the Americas, comparable technology (metates and manos) were often worked within a boxed trough thus preserving grain that would otherwise have been lost (Adams 1999).

However, the bottom line is that our studies suggest that the grinding hours needed to sustain a person in a traditional seed based economy may have sometimes been overstated.

Conclusion Four specific issues were identified for investigation in this paper, namely, millstone size, grinding action, muller type and seed variety. By establishing a number of fundamental features of grindstones, grinding procedures and seeds, the experiments allowed firm conclusions to be drawn concerning the first and last of these issues and provided worthwhile preliminary data about the other two.

67 In assessing production, the size of the grinding unit is most critical. For grooved or troughed sandstone millstones, there is a more or less linear increase in efficiency as the length (more specifically, the area) of the depression increases. Critical thresholds could not be precisely established but functional difficulties were beginning to intrude with the smallest units whilst the upper threshold is likely a biomechanical issue related to the physical stature of the grinder. The experiments demonstrate clearly that large ‘millstones’ perform considerably more effectively than small grindstones and support their status as formal specialised implements. The large and medium millstones are capable of producing a kilogram of wet meal in 1.8 hours whereas the smaller grindstones required 4.1 hours; more than twice the time. If seeds needed to be included as a major component of diets, the benefits available from the large millstones would seem to justify the costs of acquisition or development of the technology.

The seed being processed is also a critical factor in the output of ground meal. Seed selection may effectively double labour investment in grinding. However, these mean figures conceal even greater individual variations. For example, grinding panic with a very small millstone required 8.3 hours of labour to produce a kilogram of wet meal whereas grinding millet using a large millstone required only one hour. Variations of these magnitudes may have had considerable influence on economic decisions. These results with proxy grains highlight the importance of seed type and experiments aimed at establishing the grinding characteristics of a range of native seeds are being undertaken at present.

The aspect of widely distributed small grindstones and mortars having noteworthy seed grinding capability suggests that the acquisition of specialised technology represented by the large millstone was not a prerequisite for the use of seeds. Non-specialised grindstones could have contributed to the human exploitation of various areas by supplementing diets without necessitating the adoption of a ‘seed economy’. The multi-use capability of small grindstones and mortars supports suggestions that the technological capacity to utilise seeds (and possibly their actual use) may have been both early and widespread.

The grinding action used was found to be of only limited moment. A rotary grinding action is mechanically advantageous to the grinder but these advantages are only reflected in very limited increased efficiency. The crucial factor is that a rotary action is used with a grindstone of greater functional area than an equivalent length grooved grindstone worked with a reciprocal action.

The selection of a muller can be more or less critical depending on the configuration of its paired millstone. The large dished millstones were able to be operated with a wide range of mullers whereas the very small grooved millstones were particularly sensitive to muller choice and muller action. A compatible topstone or muller becomes increasingly important as the depression deepens. A poorly matched muller is unable to grind finely and may not be able to be used at all in a deep groove thus necessitating commencement of a fresh groove. Knowledge of muller type, and grinding action, would both benefit from further, specifically targeted, experimental investigation.

The seed being ground, its method of preparation, and whether ground wet or dry, all have a bearing on output and thus the hours of labour involved to sustain a person. However, the type and size of the grindstone used most critically affects efficiency. As such, this paper demonstrates conclusively that grindstone morphology is of major importance in subsistence decisions relating to seed processing. Our results also indicate that current theorising of the role of grindstone morphology in the subsistence change in Australia, particularly in the Late Holocene, requires further thought and revision of current interpretations. It is clear that it is largely meaningless to talk about grindstone efficiency without carefully delineating the grindstone and seed parameters involved. Furthermore, characteristics such as size should

68 be featured in further considerations of the evolution of grindstone technology with regard to efficiency and degree of economic dependence on seed staples.

Acknowledgments Special thanks go to Nindethana Australian Seeds, Breeders Choice Seeds, Queensland Herbarium, Australian National Botanic Gardens and Australian National Herbarium. We also thank Mike Smith, Richard Fullagar, Birgitta Stephenson , Elspeth Hayes and two anonymous reviewers for helpful discussions and Richard Fullagar and Elspeth Hayes for the invitation to present an early version of this paper at the Australian Archaeological Association Conference in Coffs Harbour in 2013.

69 References Adams, J.L. 1994 The development of prehistoric grinding technology in the Point of Pines area, east-central Arizona. Ann Arbor: UMI Dissertation Services. Adams, J.L. 1999 Refocusing the role of food-grinding tools as correlates for subsistence strategies in the U.S. Southwest. American Antiquity 64(3):475-498. Adams, J.L. 2002 Ground stone analysis; a technological approach. Salt Lake City: University of Utah Press. Arndt, W. 1961 Indigenous sorghum as food and in myth: The Tagoman Tribe. Oceania 32(2):109-112. Balme, J. 1991 The antiquity of grinding stones in semi-arid western New South Wales. Australian Archaeology (32):2-9. Balme, J., G. Garbin and R.A. Gould 2001 Residue analysis and palaeodiet in arid Australia. Australian Archaeology (53):1-6. Bennett, K.H. 1897 Descriptive list of Australian weapons, implements, etc. from the Darling and Lachlan Rivers. Sydney: Government Printer. Bird, D.W. and R. Bliege Bird 2005 Evolutionary and ecological understandings of the economics of desert societies: comparing the Great Basin USA and the Australian Deserts. In P. Veth, M. Smith and P. Hiscock (eds), Desert peoples: archaeological perspectives, pp.81-99. Malden, Mass.: Blackwell. Brokensha, P. 1975 The Pitjantjatjara and their crafts. Sydney: Aboriginal Arts Board, Australia Council. Cane, S. 1984 Desert camps: a case study of stone artefacts and Aboriginal behaviour in the Western Desert. Unpublished PhD thesis, The Australian National University, Canberra. Cane, S. 1987 Australian Aboriginal subsistence in the Western Desert. Human Ecology 15(4):391-434. Cane, S. 1989 Australian Aboriginal seed grinding and its archaeological record: a case study from the Western Desert. In D.R. Harris and G.C. Hillman (eds), Foraging and farming: the evolution of plant exploitation, pp.99-119. Sydney: Unwin Hyman. Cleland, J.B. 1966 The ecology of the Aboriginal in South and Central Australia. In B.C. Cotton (ed.), Aboriginal Man in South and Central Australia, pp.111-158. Adelaide: W. L. Hawes, Government Printer. Cleland, J.B. and T.H. Johnston 1933 Ecology of the Aboriginals of Central Australia. Transactions of the Royal Society of South Australia 57:113-124. Cleland, J.B. and N.B. Tindale 1959 The native names and uses of plants at Haast Bluff, Central Australia. Transactions of the Royal Society of South Australia 82:123-140. Davidson, D.S. and F.D. McCarthy 1957 The distribution and chronology of some important types of stone implements in Western Australia. Anthropos 52(3 & 4):390-458. Devitt, J. 1992. Acacias: a traditional Aboriginal food source in central Australia. In A. P. N. House and C. E. Harwood (eds) Australian dry-zone acacias for human food: proceedings of a workshop held at Glen Helen, Northern Territory, Australia, 7-10 August 1991. pp. 37-53.Melbourne: CSIRO Publications. Dillon, S.L., P.K. Lawrence, R.J. Henry, L. Ross, H.J. Price and J.S. Johnston 2004 Sorghum laxiflorum and S. macrospermum, the Australian native species most closely related to the cultivated S. bicolor based on ITS1 and ndhF sequence analysis of 25 sorghum species. Plant Systematics and Evolution 249(3):233-246. Fullagar, R. and J. Field 1997 Pleistocene seed-grinding implements from the Australian arid zone. Antiquity 71(272):300-307. Fullagar, R., J. Field and L. Kealhofer 2008 Grinding stones and seeds of change: starch and phytoliths as evidence of plant food processing. In Y.M. Rowan and J.R. Ebeling (eds), New approaches to old stones: recent studies of ground stone artifacts, pp.159-172. London: Equinox.

70 Fullagar, R., E. Hayes, B. Stephenson, J. Field, C. Matheson, N. Stern and K. Fitzsimmons 2015 Evidence for Pleistocene seed grinding at Lake Mungo, south-eastern Australia. Archaeology in Oceania 50:3-19. Gorecki, P., M. Grant, S. O'Connor and P. Veth 1997 The morphology, function and antiquity of Australian grinding implements. Archaeology in Oceania 32(2):141-150. Gould, R.A., D.A. Koster and A.H. Sontz 1971 The lithic assemblage of the Western Desert Aborigines of Australia. American Antiquity 36(2):149-169. Gregory, A.C. 1887 Memoranda on the Australian Aborigines (In H. Ling Roth). Journal of the Royal Anthropological Institute 16:102-135. Harney, W.E. 1951 Australian Aboriginal cooking methods. Mankind 4(6):242-246. Hayes, E. H. 2015 What was ground? A functional analysis of grinding stones from Madjedbebe and Lake Mungo, Australia. Unpublished PhD thesis, The University of Woollongong, Woollongong. Horne, G. and G. Aiston 1924 Savage life in central Australia. London : Macmillan and Co., Limited. Isaacs, J. 1987 Bush food : Aboriginal food and herbal medicine , McMahons Point, N.S.W.: McMahons Point, N.S.W. : Weldons. Kraybill, N. 1977 Pre-agricultural tools for the preparation of foods in the Old World. In C.A. Reed (ed.), Origins of agriculture, pp.485-521. The Hague: Mouton Publishers. Mauldin, R. 1993 The relationship between ground stone and agricultural intensification in western New Mexico. Kiva 58(3):317-330. McCarthy, F.D. 1976 Australian aboriginal stone implements: including bone, shell and teeth implements. Sydney: Australian Museum Trust. McCarthy, F.D., E. Bramwell and H.V. Noone 1946 The stone implements of Australia. The Australian Museum, Sydney, Memoir IX:1-94. Mulvaney, K. 1998 The technology and Aboriginal association of a sandstone quarry near Helen Springs, Northern Territory. In R. Fullagar (ed.), A closer look: recent Australian studies of stone tools, pp.74-93. Sydney: Sydney University Archaeological Methods Series. Newland, S. 1920-21 The annual address of the president. Proceedings of the Royal Geographical Society of Australasia (South Australia Branch):1-15. O’Connell, J.F. 1977 Aspects of variation in central Australian lithic assemblages. In R.V.S. Wright (ed.), Stone tools as cultural markers: change, evolution and complexity, pp.269-281. Canberra: : Australian Institute of Aboriginal Studies. O'Connell, J.F. and K. Hawkes 1981 Alyawara plant use and optimal foraging theory. In B. Winterhalder and E.A. Smith (eds), Hunter-gatherer foraging strategies: ethnographic and archaeological analyses, pp.99-125. Chicago: University of Chicago Press. O'Connell, J.F., P. Latz and P. Barnett 1983 Traditional and modern plant use among the Alyawara of Central Australia. Economic Botany 37(1):80-109. Richards, H. C. 1911. The building stones of St. John’s Cathedral, Brisbane. Proceedings of the Royal Society of Queensland. August 26th, 1911. Roth, W.E. 1897 Ethnological studies among the North-West-Central Queensland Aborigines. Brisbane: Brisbane : Govt. Printer. Smith, M., E. Hayes and B. Stephenson 2015 Mapping a millstone: the dynamics of use- wear and residues on a Central Australian seed-grinding implement. Australian Archaeology 80:70-79. Smith, M.A. 1985 A morphological comparison of Central Australian seedgrinding implements and Australian Pleistocene-age grindstones. The Beagle, Occasional Papers of the Northern Territory Museum of Arts and Sciences 2(1):23-38. Smith, M.A. 1986 The antiquity of seedgrinding in arid Australia. Archaeology in Oceania 21(1):29-39. Smith, M.A. 1989 Seed gathering in inland Australia: current evidence from seed-grinders on the antiquity of the ethnohistorical pattern of exploitation. In D.R. Harris and G.C.

71 Hillman (eds), Foraging and farming: the evolution of plant exploitation, pp.. 305-317. Sydney: Unwin Hyman. Smith, M. A. 2013. The archaeology of Australia’s deserts. Cambridge: Cambridge University Press.Smith, M.A. 2015a What sort of seed grinding at Pleistocene Lake Mungo? Archaeology in Oceania 50(3):175-176. Smith, M. A. 2015b. Western Desert tjiwa- and tjungari- type grindstones and their archaeological significiance. Australian Aboriginal Studies. 1:115-121. Spencer, B. and F.J. Gillen 1899 The native tribes of Central Australia. London: Macmillan. Spencer, B. and F.J. Gillen 1912 Across Australia. London: Macmillan. Tindale, N.B. 1959 Ecology of primitive Aboriginal man in Australia. In A. Keast, R.L. Crocker and C.S. Christian (eds), Biogeography and ecology in Australia, pp.36-51. Den Haag: Junk. Tindale, N.B. 1977 Adaptive significance of the panara or grass seed culture of Australia. In R.V.S. Wright (ed.), Stone tools as cultural markers: change, evolution and complexity, pp.345-349. Canberra: A.I.A.S. Veth, P. and S. O'Connor 1996 A preliminary analysis of basal grindstones from the Carnarvon Range, Little Sandy Desert. Australian Archaeology (43):20-22. Warner, W.L. 1937 A black civilization: a social study of an Australian tribe. New York: Harper & Brother.

4 Chapter 4 - Productivity of Native Seeds when Ground

4.1 Introduction

The third paper investigates the grinding profiles of a number of native seeds and the domesticated seeds processed in Chapter 3, using a single replicated millstone and muller pair. It assesses the various seeds for hardness, ease of grinding and fineness of the product and establishes grinding times for individual seeds and various groups such as acacias, herbs and grasses.

4.2 Methodology

An experimental methodology was adopted. A total of 150 experimental grindings were performed on the fruits of 15 native and three domesticated seed producing plant species. The seeds were ground in small batches, usually in their clean or raw state, but some were pre-treated by soaking or parching. The grinding characteristics were evaluated and the time to complete the grinding of each lot was recorded.

4.3 Conclusion in Summary

Experimental grinding of a range of both native and domesticated seeds has demonstrated conclusively that the type of seed, and some pre-treatments, are of major significance to seed processing times.

Soft ‘grass type’ seeds are usually easy and quick to grind and do not necessitate pre-treatment. However, the grinding responses of hard seeds from, for example, Acacia shrubs and trees, vary widely and need to be assessed individually as does the need for, and response to, pre-treatments.

Domestic seeds can sometimes be used as viable proxies for unavailable native seeds but more investigation is required.

4.4 The Third Paper

The paper was published as: John Mildwaters and Chris Clarkson. 2020. An experimental assessment of the grinding characteristics of some native seeds

used by Aboriginal Australians. Journal of Archaeological Science: Reports. Vol. 30. Article 102127. pp.1-15. The paper, prior to final typesetting, is as follows:

An experimental assessment of the grinding characteristics of some native seeds used by Aboriginal Australians.

Authors: John Mildwaters and Chris Clarkson.

Abstract Numerous plant seeds (the caryopsis or achene) from a wide variety of genera were traditionally ground for food by hunter-gatherer peoples, including over 200 varieties by Aboriginal Australians. In Australia, these seeds varied greatly in size, shape and hardness. Except for a broad distinction being made between grass and tree seeds, differences in the morphological and other characteristics of seeds — in particular their performance in seed- grinding — are rarely documented. A recent experimental examination of the efficiency of millstones using commercial proxies for difficult to obtain Australian native seeds has shown significant differences in workability and output of different types of seeds. This paper tests predictions as to whether the grinding characteristics of a number of widely used native seeds also vary significantly, the implications this has for the selection, treatment and use of seeds by Aboriginal people in Australia and elsewhere, and generates quantitative data which can supplement and explicate other experimental research. It also makes a preliminary assessment of whether domesticated seeds can provide useful grinding analogues for unavailable native seeds. Our findings have broad relevance for an understanding of prehistoric seed-grinding and species selection in Australia and other parts of the world.

Keywords: Seeds; Seed grinding; Processing characteristics; Productivity; Australia; Aboriginal. Declaration of Interest: None

Introduction

Aboriginal use of seed-foods was known and commented on by early explorers of inland areas of Australia (e.g. Mitchell 1839, 1848; Sturt 1849) and later in the central and northern areas by ethnographers such as Spencer and Gillen (1899). The mid-twentieth century saw archaeological consideration of hunter-gatherer economies and in particular the importance of seeds in traditional Australian Aboriginal societies (e.g. Cleland 1957, 1966; Cleland and Tindale 1954; Petersen 1978; Tindale 1972, 1977). In Australia, grass seeds such as Panicum decompositum (native millet) were seen to predominate in the east (Allen 1974) and remained of significance in other areas while Acacia and other tree and shrub seeds were of greater prominence in central and northern areas (e.g. Cane 1984; Petersen 1978). However, despite their nutritional qualities — typically 50-70 per cent carbohydrate and 10- 20 per cent protein (Smith 2013:197) — by the middle of the century widespread European disruption of traditional landuse and economy meant little use was being made of native seeds even by Aboriginal groups maintaining some form of traditional economy (Petersen 1978). By this time, European white wheat flour was in wide use and efforts were made to explain the desire for commercial flour over native seed meal (Petersen 1978), sometimes drawing on optimal foraging theory for their models (e.g. O’Connell and Hawkes 1981).

Various commentators have reported the number of edible seeds employed in traditional Aboriginal economies (e.g. Cane 1987:397). We have identified 222 native seeds that were being ground in ethnographic times, clearly illustrating the breadth and complexity of native seed use in Australia. It is important to note that the fruits of various trees, shrubs, succulents, etc. as well as grasses, were exploited (Smith 2013:198). Almost all the fruits used throughout the continent required preparation of some kind. The preparation of the seed types discussed in this paper (that is, seeds no larger than those of Brachychiton sp.) usually included grinding on a grindstone (Smith 2013:197).

Despite the wide variety in fruit types, possible differences in grinding times due to individual characteristics of the seeds being processed by indigenous gatherers has attracted little attention in the archaeological or ethnographic literature. This is in contrast to domesticated seeds, especially the cereals commonly milled for flour, which have been widely investigated in the food industry literature (e.g. Belderok, et al. 2000). For example, O’Connell and Hawkes (1981) in their influential paper on Alyawara plant use in central Australia, described the processing of Acacia aneura (mulga) seeds from collection through to grinding. Despite the extensive morphological disparities between various types of Acacia seeds, they assumed that most ‘other acacias should have seed-handling times comparable’ (1981:125). In America, a similar assumption was made by Simms who suggested that plant data he had collected were ‘best seen as being analogous for groups of plant types with similar morphological characteristics’ (1987:38). An important exception, and also the only detailed Australian study, is Cane’s work (e.g. 1989) which reported the time taken to grind quantities of a number of different types of soft seeds. In addition, a recent experimental examination of the efficiency of various millstones has shown significant differences in workability and output for the proxy commercial seeds employed (Mildwaters and Clarkson: 2018).

High-level archaeological theory building, particularly that associated with Human Behavioural Ecology, was also influenced by Australian reports which discussed seed grinding at some level (e.g. Bright et al. 2002; Edwards and O’Connell 1995; O’Connell and Hawkes 1981; Simms 1987). Unfortunately, these studies often contained limited data and made few allowances for differences between the grinding implements used or the seeds being processed. With such little data available, it is unclear how broadly applicable such limited ethnoarchaeological observations about handling times and daily yields truly are.

Smith suggests that a broad-based research strategy is necessary for archaeology to address the issues surrounding seed-grinding. Site selection, sampling strategies, grindstone morphology, macro and micro wear analysis, residue analysis and carbonised- plant recovery are all issues which warrant additional attention (2004:173-4). Seed-based economies were not necessarily monolithic and the:

use of seeds may have been regionally patchy, localised in different parts of the socioeconomic system, or involved an emphasis on different classes of seeds (tree seeds, grasses, chenopods, succulents) in different regions or at different times, or involved different technological pathways (Smith 2004:183).

Accordingly, simple presence / absence reductionist approaches noting only whether or not a particular seed or grindstone were recovered from a site do little to illuminate a complex whole (2004:182).

Given that Aboriginal economies would vary over time and according to environment, it is necessary to compare differences at local and then regional levels. To investigate these differences, detailed information on specific seed characteristics, including their yields when ground, is needed, because at present there are ‘few quantitative data on the collecting and processing of seeds’ (O’Connell et al. 1983: 90). Even available data need to be treated with

76 some caution as detailed seed processing knowledge, particularly of tree seeds, may have already been unavailable at the time of recording (Cane 1987:400). Experimentation therefore remains one of the few avenues available to acquire quantitative comparative information about the processing qualities of a wide range of seeds, although detailed and systematic observation of contemporary Aboriginal seed grinding remains an urgent and extremely valuable undertaking.

In an endeavour to identify and isolate productive performance differences between seeds when being ground, this paper expands on the work of Mildwaters and Clarkson (2018), which performed a number of experiments grinding commercially available domesticated proxies for native seeds on a wide range of replica Australian Aboriginal grindstones. In this paper we determine the milling characteristics of 15 species of commercially obtained undomesticated native seeds plus three domesticated seeds, by way of experimental grinding using a replicated millstone and muller. Three of the seeds are commercial varieties available world-wide and many of the remainder, while native to Australia, have morphologically comparable relatives elsewhere. By establishing the times required to grind various seeds, the experiments provide basic quantitative information on various aspects of the grinding performance of a number of seed species.

Methods

A total of 150 experimental grindings were performed on the fruits of 18 seed producing plant species. The seeds were ground in small batches, usually in their clean or raw state, but some were pre-treated (see later). The grinding characteristics were evaluated and the time to complete the grinding of each lot was recorded. We describe below the range of seeds, grindstones, pre-treatment methods and recording procedures used in the experiments.

Seeds and pre-treatments

Seeds from fifteen species could be acquired commercially in enough quantity to allow experimentation. Six varieties of seeds are from trees (five from Acacias), two from shrubs, five from grasses and two from herbs (or succulents). Few of the grass seeds were in a clean form (naked caryopsis) and accordingly required de-husking. We found de-husking to seldom be a simple procedure, such as rubbing between the hands or soaking as often depicted in ethnographic reports, thus indicating Aboriginal people had specific skills and knowledge for this task that either have not been satisfactorily recorded or that we lack. The three domestic grass seeds previously mentioned were also included to facilitate comparisons and to establish whether they could serve as proxies for native seeds not always readily available. The domesticates selected are morphologically comparable to a core range of small, medium and large sized native seeds (Mildwaters and Clarkson 2018). However, they are not physically representative of the very small seeds such as those of the various herbs or the very large seeds of some of the trees used in Australia. Many native seeds have considerable natural size variation which can be further amplified by growing and harvesting conditions (Sweedman 2006:67). The seed dimensions recorded are of typical examples taken from the samples purchased and do not vary materially from the sizes listed by Sweedman (2006).

Descriptions of the plants used in this study are presented in Table 1 and are largely summarised from Flora of Australia (Mallett and Orchard 2002) or, for the grasses, have been downloaded from http://ausgrass2.myspecies.info/ on 08/02/2018. Present-day

77 distributions of plants, the number of ethnographic reports noting their use, and a reference to an example of one such report, are included for each species in Table 1.

Botanical Name Common Names Type Status Seed Pre- Ethnographic Distribution Ethno. Example Reference treatment Reports

Acacia acuminata Raspberry jam wattle Tree Native Ovate, dimpled, shiny black, small P. SW WA to semi-arid. 5 Bird & Beeck 1988 orange aril. 4x3 mm. Acacia aneura Mulga Tree Native Ovate, shiny brown. 5x4 mm. P, S, Pc. Extensive semi-arid to arid. 35 Cleland 1966

Acacia coriacea Dogwood; River jam; Tree Native Oval,dimpled shiny black, with orange P, S. Northern Australia semi-arid to 12 Golson 1971 Wirewood; Desert oak. aril. 7x4 mm. arid. Acacia cowleana Halls Creek wattle Tree Native Flat oval, shiny black with small orange na Central to Northern Australia 3 Isaacs 1987 aril. 4 x 1.5 mm. semi-arid to arid. Acacia victoriae Elegant, Prickly wattle Tree Native Spherical, shiny mottled brown-black. P, S, Pc. Widespread drier areas. 10 Gason 1886 4.5 mm dia.. Astrebla lappacea Curly Mitchell grass Grass Native Oval/ Ovate, orange. 2x1 mm. na Eastern to Central drier areas. 7 Turner 1895

Atriplex nummularia Old man, Bluegreen, Giant Shrub Native Oval Pink. 2x1 mm in 4 mm disced, P. Central to Southern drier areas. 1 Isaacs 1987 saltbush spongy flower Brachychiton acuminatus Rock kurrajong; Bottletree Tree Native Ovate, mottled brown. 8x5 mm. P, S, Pc. North to west Australia 2 Tindale 1972

Eragrostis eriopoda Woollybutt; Naked Grass Native Oval, red. 1x0.5 mm. na Widespread central and western 25 O'Connell et al. 1983 wollybutt; Neverfail Australia Panicum decompositum Native millet; Papa; Grass Native Lenticular, brown. 2x1 mm. na Australia wide 32 Latz 1995 Umbrella grass Panicum maximum Gatton panic; Green panic Grass Domestic Lenticular, brown. 2x1 mm. na Global n.a n.a

Panicum miliaceum White French millet Grass Domestic Lenticular, yellow-white. 2.5x2 mm. S, Global n.a n.a

Portulaca oleracea Pigweed; Purslane; Herb Native Spherical, black. 0.5 mm dia. na Australia wide. 49 Cane 1987 Munyeroo Sida rohlenae Shrub sida Shrub Native Wedge shaped, spined pericarp, red 3x2 P. Northern and Central Australia 2 O'Connell et al. 1983 mm. Cariopsis 2x1mm.

79 Sorghum bicolor Red sorghum Grass Domestic Spherical, Red. 4 mm dia. Pc, Global n.a n.a

Sorghum plumosum Plume canegrass Grass Native Lanceolate, brown. 3x0.5 mm. na Wetter northern Australia. 1 Arndt 1961

Tecticornia verrucosa Samphire, Swamp grass Herb Native Spherical, black. 0.5 mm dia. variable. na Central and western Australia 6 Smith 1986

Triodia pungens Soft, Gummy spinifex Grass Native Oval, yellow. 2x1 mm na Arid Australia 3 Cane 1987

Table 1 - Details of native and domesticated seeds tested. Pre-treatments- P=Parched; S= Soaked; Pc= Pre-cracked.

80 Aboriginal people sometimes pre-treated various seeds by parching, soaking or cracking to improve grinding qualities (e.g. Devitt 1992; Kimber and Smith 1987; Smith 1986). The reason for parching is generally given simply as ‘to improve grinding’ (e.g. Cane 1987:401), but this may also include the removal of oils which encourage deposits that clog abrasive surfaces (see below). Cracking furthered the grinding process but may have also ensured that hard seeds were not lost by bouncing off the millstone at the initial attempt to grind, as was occasionally experienced in our experiments. Soaked seeds were sometimes ground wet (e.g. Bindon 1996:24; Roth 1901:12) but, in Australia, appear to have been more commonly sun dried before grinding (e.g. Cane 1987:402; Roth 1897:s107).

Multiple pre-treatments, for example, parching and then cracking, were considered to be beyond the ambit of these experiments and were not attempted. However, as parching or soaking in bulk required little time investment, the possible benefits of compound pre- processing should be kept in mind when assessing results, particularly those of widely exploited seeds which did not perform well in the experiments.

Experimental preparation of conditioned grain samples involved the following steps - 1. parching in an oven for 15 minutes at 200 degrees Celsius; or 2. soaking in water for four hours, based on Cane’s (1987:401) comment that seeds were ‘soaked for several hours’, and then sun dried for at least one hour; or 3. cracking in a shallow sandstone mortar with a water-worn pebble pestle (see Mildwaters and Clarkson 2018 for a description of these implements).

Table 2 lists the terms used in the analysis to describe the ‘condition’ of the seeds prior to experimental grinding.

Condition Code Comments Clean C Husks, arils etc. removed. Naked caryopsis Raw R Not entirely ‘clean’. May be in a fruit or case or have an aril attached etc. Parched P A clean or raw seed heated in an oven for 15 minutes at 200° C. Soaked S A clean or raw seed soaked in water for 4 hours than dried. Cracked Pc Seeds pre-cracked, usually with a mortar and pestle Cracking na Initial cracking of the seeds separate from grinding. Table 2 - Description of seed conditions used in the experiments.

Those examples utilising pre-treatments increased the number of individual experimental variations to 36. However, pre-cracking the grains was found to constitute a material portion of the grinding task and thus distorted comparison times. These examples have therefore been excluded from analysis unless otherwise stated. A number of other grindings were also excluded for various reasons leaving 125 grinding episodes for analysis.

All seeds were acquired from commercial seed suppliers with the majority being purchased from Nindethana Seed Service Pty Ltd in Western Australia. The seeds used in the experiment are briefly described below and are shown in Figure 1 and Figure 2.

A B C

D E F

G H I

J K L

M N O

P Q R

S T

82 Figure 1 – Seeds and meal produced by grinding in the experiments photographed against a 2 mm grid.

Panicum maximum - green panic: A. De-husked (clean) seeds but some retaining an inner glume. B. Ground meal in which some glumes can still be discerned; Panicum miliaceum - white French millet: C. Clean seeds. D. Finely ground meal or flour; Sorghum bicolor - red sorghum: E. Clean seeds. F. Ground meal; Acacia acuminata - raspberry jam wattle: G. Clean seeds. H. Ground meal. I. Seed cases clogging underside of muller; Acacia aneura – mulga: J. Clean seeds. K. Ground meal. L. Ground and partially ground mulga seeds with millstone and muller which both show heavy waxy reflective fouling after grinding; Acacia coriacea – wirewood: M. Clean seeds. N. Ground meal; Acacia cowleana - Halls Creek wattle: O. Clean seeds. P. Ground meal; Acacia victoriae – elegant wattle: Q. Clean seeds. R. Ground meal; Astrebla lappacea - Curly Mitchell grass: S. Clean seeds. T. Ground meal. (Photographs by John Mildwaters)

A B C

D E F

G H I

J K L

M N O

P Q R

S T U

84 Figure 2 – Remaining seeds and meal produced by grinding in the experiments photographed against a 2 mm grid. Atriplex nummularia - old man saltbush: A. Spongy fruits. B. Ground meal showing flattened flakes; Brachychiton acuminatus - rock kurrajong: C. Clean seeds. D. Ground meal; Eragrostis eriopoda - woollybutt grass: E. Clean seeds. F. Ground meal; Panicum decompositum - native millet: G. Clean seed (note that inner glumes have been removed from some by the ‘flash’ husking H. Clean green panic (top) and native millet (bottom). I. Ground meal - green panic (left), native millet (right); Portulaca oleracea – pigweed: J. Clean seeds. K. Ground meal; Sida rohlenae - shrub sida: L. Spined and cased seeds. M. Partially husked seeds. N. Rough ground meal; Sorghum plumosum - plume canegrass: O. Cased seeds with awns removed. P. Ground meal; Tecticornia verrucosa - samphire: Q. Clean seeds. R. Ground meal; Triodia pungens - soft spinifex: S. Partially husked seeds. T. De-husked seeds. U. Ground meal. (Photographs by John Mildwaters)

The Seeds

Domesticated seeds

Panicum maximum (green panic) is a small, soft, spongy domesticated seed morphologically similar to the widely exploited Panicum decompositum or native millet (Figure 1A). It retains some inner glumes even after commercial de-husking (Latz 1995).

Panicum miliaceum (white French or common millet) is a hard, sharp and brisant (shattering) domesticated grain (Figure 1C). It is one of the larger domesticated millets but is comparable to the slightly smaller and more elliptical (but unavailable) native, Brachiaria miliiformis (arm grass millet). O’Connell et al. stated that the native millets are ‘morphologically and taxonomically similar to those domesticated’ (1983:99). Without knowing exactly which millets were being discussed, it is reasonable to expect that Panicum miliaceum is representative of the larger seeded native grasses.

Sorghum bicolor (red sorghum) is closely related to the native sorghums (see Sorghum plumosum) but the seeds are larger and roundish — see Figure 1E. The seeds have a hard, resilient texture and are initially difficult to crack. This seed was included in the experiments as it is morphologically similar to a number of Acacia species (Mildwaters and Clarkson 2018).

Native Seeds

Acacia acuminata (raspberry jam wattle) was widely utilised in Western Australia. It is a relatively small seed of firm consistency (Figure 1G). In the central north, it appears to have been ground with a conventional millstone but in the south, at least on occasions, it was processed with the little mentioned ‘dimpled grindstone’, which is presumably a type of mortar:

85 'A small quantity of seeds is then placed in the hollow in the [dimpled] grindstone and smashed with a second stone to break the hard outer casing. The inner material is ground' (Bird and Beeck 1988:119).

Acacia aneura (mulga) is widely distributed and has a hard, rubbery, relatively large oval ‘clean’ seed sometimes with an oily finish (Figure 1J). It is comparable in size to domesticated Sorghum bicolor. It was a practically ubiquitous Aboriginal food in the drier areas of the continent and its ‘combination of high productivity with wide availability’ made it ‘a singularly valuable plant within the traditional inventory’ (Devitt 1992:38). According to Smith (1986), in his research area in Central Australia, Acacias were pounded in a mortar before grinding. The seed was usually parched before grinding (Devitt 1992:51; Goddard and Kalotas 2002:35; Smith 2013:197). Mulga was the grain upon which O’Connell and Hawkes (1981:124) initially based their influential ethnographic seed grinding output results of 3.94 hours per kilogram.

Acacia coriacea (wirewood) (Figure 1M) is known by various names in different areas. It was widely exploited and considered of major importance in some areas (Cane 1987:398; Devitt 1992:38). Unlike most Acacias, before ripening the seeds could be ‘eaten raw like green peas’ (Devitt 1988:141; Isaacs 1987:112; O’Connell and Hawkes 1981:125) thus enhancing their ranking in foraging models. The seed exudes traces of oil when stored. However, the seeds needed to be ground when ripe and, as they become ‘very hard, grinding may be preceded by a pounding step’ (O’Connell and Hawkes 1981:125). O’Connell and Hawkes (1981:125) assumed that A. coriacea would have a similar grinding rate of 3.94 h/kg to A. aneura but warned about the possibility of needing pre-crushing.

Acacia cowleana (Halls Creek wattle) (Figure 1O) is not frequently mentioned, however reports of its seeds being ground exist for Central Australia and northern Western Australia (Bindon 1996:10; Isaacs 1987:111; O’Connell and Hawkes 1981; O’Connell et al. 1983:86). O’Connell and Hawkes (1981:125) assumed its grinding profile to be similar to mulga— 3.94 hrs/kg.

Acacia victoriae (elegant wattle, prickly wattle or bardi bush) has a relatively large clean seed with a slightly oily feel (Figure 1Q). It is comparable in size to domesticated sorghum. It has a wide distribution in the drier areas of the continent and reports of it being ground for food are common (e.g. Bindon 1996:31; Isaacs 1987:234; Smith 1985:29). It is regarded as a hard seed which requires cracking as a minimum prior to grinding (Devitt 1992:51; Smith 1986:31). Kimber and Smith provide a detailed list of processing steps:

Very hard seeds such as those from prickly wattle [Acacia victoriae] require a long sequence of roasting, cleaning, pounding, cleaning, pounding again, dry grinding in a mortar, and finally wet grinding with water on a millstone into a paste before being eaten (1987:234).

Astrebla lappacea (curly Mitchell grass) is one of a number of tussock Mitchell grasses and the use patterns of the various species are not clear. According to Flora of Australia the ‘nomenclature has been very confused and in error in the past’ (Mallett 2005:453). The literature overall contains only a handful of reports on the use of the Mitchell grasses and distinctions between the main species (A. lappacea, A. pectinata, A. elymoides and A. squarrosa) are seldom made clear. Most of these early comments relate to the more eastern dry areas (e.g. Turner 1895:13) and these frequently identify plants only to genus level or with A. pectinata seemingly adopted as a general name for all the species (Golson 1971:197). In essence, it would seem that the various species have morphologically different seed heads but similar seeds. Accordingly, for the purposes of these experiments, no attempt has been made to distinguish between the species.

86 The species obtained, A. lappacea, has seeds which are orange in colour (Figure 1S). They are not visually distinguishable from Astrebla pectinata seeds except by colour (the seeds of A. pectinata are a brighter orange in the samples on hand) and can reasonably be accepted as a substitute for the other Astrebla types. The seeds are also visually similar to the widely used Panicum decompositum (native millet).

A possible reason for the paucity of use reports was that the Mitchell grasses may not have been a preferred choice for grinding due to their being very difficult to de-husk (Dr. Ian Chivers, Native Seeds Pty Ltd, Pers. Comm.). This was confirmed as the only sample able to be obtained was in floret and spikelet form. Manual husking proved challenging and the de-husked (clean) grains still retained minor residual husks.

Atriplex nummularia (old man saltbush) is a member of a large family (Atriplex and certain members of the genus Chenopodium) and the extent and details of individual saltbush grindings for food are not entirely clear. However, it is certain that some use was made of the seeds and the sample obtained, old man saltbush, was widely distributed (Allen 1974: 312; Isaacs 1987:115). Raw old man saltbush seed is contained within a spongy, dry orbicular fruit (Figure 2A) and the caryopsis is extremely difficult to remove (Bonney 1994:81).

Brachychiton acuminatus, the rock or northern desert kurrajong (a pachycaul or bottle tree) is concentrated in the north of Western Australia in central and western areas although examples from further east are known. Its seeds are not specifically recorded as being ground although a number of Western Australian reports from its most populous areas identify ‘kurrajongs’ only to species level and confusion is possible with either one or both of the similarly named desert kurrajong or the northern kurrajong. Tindale (1972:234), in describing the collection of seeds for ‘pounding’ from ‘the dung and coughings of ravens’ appears to be referring to kurrajongs in general and Isaacs indicated that the various subspecies of Brachychiton ‘grow all over the country and most of the seeds were once gathered and eaten’ (1987:88).

Sufficient samples of the more commonly reported tree seeds such as desert kurrajong (Brachychiton gregorii) and northern kurrajong (B. diversifolius) were unable to be obtained. However, the large, clean seeds of rock and desert kurrajong are morphologically very similar as is their chemical composition with their protein content the same at 22% (Rao 1991) (Figure 2C). It seems reasonable to assume that, for a preliminary assessment, grinding characteristics are comparable.

Eragrostis eriopoda (woollybutt grass) is a very small seeded, but prolific producing, love grass (Figure 2E). It has a wide distribution, being found in all but the wetter areas of the continent. It can flower all year and was a popular grass for grinding. Thirteen accounts of its being processed have been located in the literature. It is soft, relatively easy to husk and was a staple in central Australia (Smith 1986:29).

Panicum decompositum (and various other Panicum grasses) have a virtually continental wide distribution in Australia and native millet (Figure 2G) is rare only in southern Western Australia and Tasmania. It is ‘morphologically and taxonomically similar to’ the domesticated millets (O’Connell et al. 1983:199). While its seed is small, it was one of the most regularly ground seeds and was a staple in central Australia (Smith 1986:29). It was also one of the earliest species of seed reported as being ground and at least 30 ethnographic reports of grinding exist (e.g. Gason 1886:79; Mitchell 1848:89). Native millet has a soft, but somewhat spongy seed. The sample obtained was in husked form and manual de-husking detached slightly more of the inner glumes than are removed from the similar commercial Panicum maximum.

87 Portulaca oleracea is a traditionally popular succulent pigweed with a black and very small seed (Figure 2J) which was considered easy to collect and process (Cane 1984:84). It was a staple in central Australia (Smith 1986:29).

Sida rohlenae (shrub sida) is widely distributed across central and northern Australia. The seeds of a number of sida species were ground but usage and the differences between them are not clear. The raw seeds (or more correctly, pericarp, flowers or burrs) of shrub sida are red-brown to black, spined, spongy and wedge shaped (Figure 2L). The sample obtained was in this form.

Sorghum plumosum (plume canegrass) (Figure 2O and Figure 2P) is one of the wild sorghums widely distributed throughout northern Australia in the wet and semi-wet areas. These sorghums are closely related to the domesticated species Sorghum bicolor (Dillon et al 2004), however the seed is much smaller and elongated with a long, springy awn (tail).

There are a small number of mentions of wild sorghums being ground for flour but the only detailed report is relatively late (Arndt 1961:109; Smith 1986:31). The accepted botanical name for Sorghum plumosum has changed regularly over time and a number of synonyms are known. This difficulty in precise identification may be a factor in the scarcity of use recordings. However, of more likely import is the fact that its distribution coincides with major seed grinding areas mainly on its southern fringe.

Seed sufficient only for a single one-gram load was able to be obtained but has been included for interest, as samples of native grass seeds are so difficult to obtain. Because of the small amount available, this seed was not included in statistical tests.

Tecticornia verrucosa, ‘this ‘peculiar samphire’ as Latz (1995: 55) described it, is a succulent in the Chenopodiaceae family (Cane 1989; Smith 2013). Its seeds are of variable morphology but predominately are soft, round, black and tiny (Figure 2Q). It was a staple in central Australia (Smith 1986:29). According to Cane it ‘grew on saline claypans and its edible seeds had to be soaked in water before it was ground’ (1989: 110).

Triodia pungens (soft spinifex) is one of the numerous Triodia species of hummock grasses which are found Australia wide on the mainland and often, incorrectly, referred to as types of spinifex. Soft spinifex (using the misapplied but common name) is found across central and northern areas of the continent. Numerous reports confirm a number of spinifex types including T. pungens were ground for flour but precise details are not clear (e.g. Cane 1984:86). The soft spinifex sample obtained was still partially husked (Figure 2S). Further manual de-husking was difficult and required great care not to pulverise the endosperm. However, a fully clean seed was still not obtained (Figure 2T).

In traditional settings, additional processing was sometimes necessary after grinding, such as winnowing or yandying (bumping or vibrating on a surface) to remove unwanted parts such as black seed cases on Acacias. No attempt was made in this study to further refine flour beyond the basic ground meal although in some cases little extra effort would have been required. With Acacia victoriae, for example, the seed cases easily separate out on a flat surface if gently ‘yandied’. Experimental grindstone and muller

A millstone and muller were created using medium grain sandstone from Helidon, Queensland (see Mildwaters and Clarkson 2018 for details of this raw material). The millstone measured 63 by 48 by 4 cm and its surface was uniformed with a carborundum block. A functional area of approximately 33 by 18 cm (594 cm²) was delineated on this surface. No groove was created. The muller fashioned was a flat slab 16 by 10 by 4 cm with

88 its trailing edge ground to a curved bevel and with a functional surface of around 165 cm² (see Figure 3). The combination is considered representative of an unused ‘large seed- grinder’ and a well-used muller pair and generally comparable with the ethnographic examples discussed by Mildwaters (2018).

Figure 3 – Photograph of the grinding surfaces of the replicated large millstone and muller used in the experiments. (Photograph by John Mildwaters)

Experimental procedures

Extensive experimentation shows that the maximum individual seed load that can be comfortably ground on such a millstone and muller combination is around five grams (depending on the seed type), and this weight was selected as the maximum load for the experiments. The minimum load adopted was one gram. This was because of the scarcity of some seeds, but also because, with very large seeds, one gram represented only a few individual grains. An intermediate load was set at either two or three grams.

Consistent with ethnographic descriptions, the rounded edge of the muller was typically utilised with a short, rocking motion to initially crack the grains (Mildwaters and Clarkson 2018). Then, by alternately raising the proximal and distal edges with long strokes, the seed was swept under the muller and ground.

Following the methods outlined in Mildwaters and Clarkson (2018), the seconds taken to grind each load were timed with an electronic stop watch and all weights were measured with electronic scales to 0.01g. Data were recorded in a Microsoft Excel spreadsheet and analysis was undertaken using Excel and SPSS Statistics. A detailed photographic record was maintained, with the seeds and the meal photographed against a two-millimetre grid.

Apart from some wet grindings conducted mainly as checks, grinding was performed dry for practical reasons described below. There is a general perception that wet grinding was prevalent traditionally but many ethnographic examples of dry grinding are also known (e.g. Horne and Aiston 1924:54; Newland 1921:14; Roth 1897:s107). For experimental purposes involving small quantities, wet grinding inhibits the generation of accurate output results because the product of wet grinding sticks to all surfaces thus increasing losses in transfers between implements. Wet grinding was considered inappropriate in view of the limited availability of some grains.

Subjective assessments for grinding qualities were recorded on a five-point rank scale. ‘Hardness’ evaluated the actual difficulty in fracturing the seed, ‘Ease’ assessed the effort involved in grinding, and ‘Fineness’ considered the caking qualities and the fineness of the

89 meal. Averaging can result in fractional values. Table 3 provides further information on how the codes were applied.

Code 1 2 3 4 5 Hardness Hard; Tough; Firm; Stiff; Resilient; Pliant; Soft; Unyielding; Resistant Pliable; Readily Yielding; Difficult to Rubbery fractured Flexible break Ease Unsatisfactory; Acceptable; Typical; Good; Crisp; Excellent; Will not grind. Hard; Spongy; Effort Moderate Soft; Sharp; May foul. May cause required; effort ; No Minimal fouling Some problems effort. Easy problems Fineness Sandy; Some Gritty. Soapy; Plastic; Cakes Sticky; Fine unground seeds; Difficult to Acceptable well; Fine grind; Cakes Will not cake. cake. Only fineness; grind; Sticks readily; medium Cakes may to implements Adheres fineness disintegrate mainly to self.

Table 3 – Subjective grinding qualities of seeds assessed by Hardness, Ease of grinding and Fineness of the meal produced.

It was initially hoped to situate our evaluation of grain hardness within the food sciences literature, in particular where protein percentages for the particular seed were available. However, grain hardness is a result of a complex interaction between a number of factors of which protein content is only one (Anjun and Walker 1991; Stenvert and Kingswood 1977). In view of this, it was determined to rely on a purely subjective assessment of grain hardness.

Some hard, brisant seeds seem to have a self-cleaning or rejuvenating aspect to them in that they encourage renewal of the stone by facilitating erosion of the contact surfaces as grinding proceeds. One such seed — white French millet — was selectively ground for short periods between experimental sessions as a means of evaluating the abrasiveness of the millstone and muller to ensure grinding performance was unimpeded.

Results

The results of the grinding experiments are summarised in Table 4 and Table 5. (The pre- cracked experiments have been excluded). Table 4 shows (by seed condition), firstly, the average time in seconds to grind one or more loads of one gram (the ‘1-gram Load’ column), secondly, the average time in seconds per gram to grind all loads of various weights (the ‘All Loads’ column) and finally, the three subjective (averaged) grinding qualities applicable to the seed and condition. Most results reported hereafter are based on the ‘All Loads’ column unless stated otherwise.

In Table 5, individual seed grinding combinations are arbitrarily grouped by average grinding time per gram into: fast =  22 seconds (N = 51); moderate = 22.1 - 54.9 seconds (N = 56);

90 and slow =  55 seconds (N=18). In general, the very easy to grind grasses and herbs fell into the fast group and, in the slow group, where grinding time per gram exceeded 55 seconds, perusal of the individual grindings log revealed that problems had usually been encountered. From the experiments, we assumed that slow combinations would likely not be attractive to Aboriginal people in the past, but other factors such as cracking or multiple pre-conditioning may well have made the seeds acceptable in a traditional economy.

Table 6 shows greater detail on the effects of seed pre-treatments. It also combines and summarises the fast and moderate native seed measures and separates out those falling into the slow group. (Excluding the slow group allowed elimination of most outliers for statistical analysis).

91 Summary of Seed Grinding by Time Seed Details 1 gram Load All Loads Grinding Qualities Ave. Secs. Secs/ Hard- Fine- Rank Botanical Name Common Name Condition No. for 1g No. gram ness Ease ness 1 Portulaca oleracea Pigweed Clean 2 17 6 10 5.0 5.0 5.0 2 Tecticornia verrucosa Samphire Clean 2 20 6 12 5.0 5.0 5.0 2 Panicum miliaceum White French millet Soaked 1 18 3 12 3.0 5.0 5.0 4 Eragrostis eriopoda Woollybutt grass Clean 2 20 6 13 4.7 4.8 5.0 5 Acacia acuminata Raspberry jam wattle. Parched 1 21 3 14 4.0 4.0 5.0 5 Panicum miliaceum White French millet Clean 1 27 7 14 2.0 5.0 5.0 7 Sida rohlenae Shrub sida Clean 2 19 4 15 4.0 4.5 3.5 8 Sorghum bicolor Red sorghum Clean 2 26 6 19 1.0 3.0 4.3 9 Panicum decompositum Native millet Clean 2 28 6 20 4.3 4.5 4.2 9 Sida rohlenae Shrub sida Parched 2 28 6 20 3.0 3.5 2.2 11 Panicum maximum Green panic Clean 2 26 6 21 4.2 4.0 4.0 12 Astrebla lappacea Curly Mitchell grass Clean 1 30 3 26 5.0 4.0 4.0 13 Acacia aneura Mulga Parched 1 34 4 28 4.0 3.0 3.8 13 Acacia cowleana Halls Creek wattle. Raw 2 35 6 28 1.5 4.0 3.8 15 Triodia pungens Soft spinifex Clean 3 32 4 30 5.0 4.0 5.0 16 Brachychiton acuminatus Rock kurrajong Parched 1 36 3 34 2.0 3.0 4.0 16 Acacia aneura Mulga Soaked 1 47 3 34 3.0 3.0 3.3 18 Brachychiton acuminatus Rock kurrajong Soaked 1 53 2 36 2.0 2.0 4.0 19 Sida rohlenae Shrub sida Raw 1 67 4 37 3.0 3.3 2.8 20 Sorghum plumosum Plume canegrass Raw 1 38 1 38 5.0 4.0 5.0 20 Brachychiton acuminatus Rock kurrajong Clean 1 59 3 38 2.0 2.0 4.0 22 Sida rohlenae Shrub sida Soaked 1 60 3 39 4.0 2.7 3.0 22 Acacia aneura Mulga Clean 1 38 3 39 2.0 2.0 4.0 24 Acacia coriacea Wirewood Parched 1 84 5 54 3.0 1.6 2.8 25 Acacia victoriae Elegant wattle Parched 1 76 3 59 2.0 2.3 3.3 26 Acacia victoriae Elegant wattle Clean 1 89 3 64 1.0 2.3 3.3 27 Acacia coriacea Wirewood Soaked 1 110 3 68 1.0 2.0 4.0 28 Acacia victoriae Elegant wattle Soaked 1 120 3 69 2.0 2.3 3.3 29 Acacia acuminata Raspberry jam wattle. Raw 2 153 5 75 2.0 2.8 3.2 30 Atriplex nummularia Oldman saltbush. Raw 1 60 2 93 4.0 2.0 2.7 31 Acacia coriacea Wirewood Raw 1 116 1 116 3.0 1.0 2.0 32 Triodia pungens Soft spinifex Raw 1 120 1 120 5.0 4.0 3.5 33 Atriplex nummularia Oldman saltbush. Parched 1 300 1 300 3.0 1.0 2.0

Counts and Averages 45 61 125 48 3.2 3.2 3.8

Table 4 – Summary of experimental seed grinding by average time in seconds to (a) grind one gram and (b) the average grinding time per gram based on all loads of various weights. Grinding qualities are based on (b). Previously cracked seeds are excluded.

Seed Seed

Time Time

Rank

Name

Grinding Grinding Condition Fast GT ≤ 22 secs. 10 Portulaca oleracea Clean 1 12 Tecticornia verrucosa Clean 2 13 Eragrostis eriopoda Clean 3 14 Acacia acuminata Parched 4 15 Sida rohlenae Clean 5 20 Panicum decompositum Clean 6 20 Sida rohlenae Parched 6 Medium GT 22.1 -54.9 secs. 26 Astrebla lappacea Clean 8 28 Acacia aneura Parched 9 28 Acacia cowleana Raw 9 30 Triodia pungens Clean 11 34 Acacia aneura Soaked 12 34 Brachychiton acuminatus Parched 12 36 Brachychiton acuminatus Soaked 14 37 Sida rohlenae Raw 15 38 Brachychiton acuminatus Clean 16 38 Sorghum plumosum Raw 16 39 Acacia aneura Clean 18 39 Sida rohlenae Soaked 18 54 Acacia coriacea Parched 20 Slow GT ≥55secs. 59 Acacia victoriae Parched 21 64 Acacia victoriae Clean 22 68 Acacia coriacea Soaked 23 69 Acacia victoriae Soaked 24 75 Acacia acuminata Raw 25 93 Atriplex nummularia Raw 26 116 Acacia coriacea Raw 27 120 Triodia pungens Raw 28 300 Atriplex nummularia Parched 29 Table 5 - Summary of overall grinding times for native seeds- average seconds per gram by seed type, condition and rank.

93 Seed Total Grinding Time < 55Grinding Time >= 55 No.of No. of Seconds No. of Seconds Scientific Name Common Name Loads Condition Loads to Grind Loads to Grind Raspberry jam Raw 5 75 Acacia acuminata wattle 8 Parched 3 14 Clean 3 39 Parched 4 28 Acacia aneura Mulga 10 Soaked 3 34 Raw 1 116 Parched 5 54 Acacia coriacea Wirewood 9 Soaked 3 68 Acacia cowleana Halls Creek 6 Raw 6 28 Clean 3 64 Parched 3 59 Acacia victoriae Elegant wattle 9 Soaked 3 69 Curly Mitchell Astrebla lappacea grass 3 Clean 3 26 Raw 2 93 Atriplex nummulariaOld man saltbush 3 Parched 1 300 Clean 3 38 Brachychiton Parched 3 34 acuminatus Rock kurrajong 8 Soaked 2 36 Eragrostis eriopoda Woollybutt 6 Clean 6 13 Panicum decompositum Native millet 6 Clean 6 20 Panicum maximum Green panic 6 Clean 6 21 Clean 7 14 Panicum miliaceum White French millet 10 Soaked 3 12 Portulaca oleracea Pigweed 6 Clean 6 10 Clean 4 15 Raw 4 37 Parched 6 20 Sida rohlenae Shrub sida 17 Soaked 3 39 Sorghum bicolor Red sorghum 6 Clean 6 19 Sorghum plumosum Plume canegrass 1 Raw 1 38 Tecticornia verrucosa Samphire 6 Clean 6 12 Clean 4 30 Triodia pungens Soft spinifex 5 Raw 1 120 Number 125 103 22 Average 24 107 Table 6 - Grinding Times - < 55 seconds and ≥ 55 seconds per gram for all loads for various conditions

Processing Times for Different Plant Types

Differences in grinding performance were apparent both between and within seed-plant types, especially with shrubs and trees. Figure 4 shows the average grinding time in seconds per gram based on all loads of various weights for the four seed groups. All grinding episodes are incorporated including those elsewhere excluded as being ‘slow’.

Figure 4 - Boxplot showing grinding time for all episodes, including those in the ‘slow’ group, of Herbs (N = 12); Grasses (N = 20) Shrubs (N =10) and Trees (N =50).

Herbs

The two herbs, clean pigweed (Portulaca oleracea) and raw samphire (Tecticornia verrucosa) were the fastest to grind requiring only 10 and 12 seconds respectively. P. oleracea was identified by Cane as being one of the easiest seeds to grind (1984:84) and experimentally, his assessment was confirmed (Figure 1K). The need to soak samphire (Cane 1989) likely relates to some factor such as reducing its salt content rather than any difficulty in grinding (Figure 1R). Both herbs possess soft but sharp seeds which were very easy to grind and produced fine flour (if slightly sandy for samphire). Grasses

Our expectation was that the grasses (along with pigweed) would prove highly productive (e.g. Tindale 1977) and indeed all five native grasses recorded fast or medium grinding times. The fastest (by around 35%) was Eragrostis eriopoda (Figure 2F). Second was the widely exploited Panicum decompositum (native millet) which, while soft, proved to be a somewhat spongy (pliant), seed which processed into slightly fibrous flour as ‘the hard inner husk is not removed’ by de-husking (Latz 1995:240) (Figure 2I). It ground and caked readily

95 but not quite as easily as those seeds which are both soft and brisant. The need for manual de-husking by flash-flaming could have resulted in slight parching, however grinding the seed in parallel with the commercial green panic (Panicum maximum) revealed that both seeds ground to a fine consistency in the same amount of time (Figure 2H). Any grinding differences between the two seeds were indistinguishable by the experimental methods adopted.

The ‘clean’ seed of Astrebla lappacea (curly Mitchell grass) retained some husk particles from manual de-husking and required 26 second to grind. The flour was not fibrous like that of panicum (Figure 1T)

The Triodia pungens (soft spinifex) seeds obtained included their husks and required two minutes to grind, and then only to a stalky, fibrous mass which would not cake. With partial manual de-husking, grinding performance of the cleaned seed improved to a medium rating of 30 seconds. The moderate-slow time for a small, soft seed resulted from the remaining husks being difficult to incorporate into the meal and would no doubt have improved if a truly clean seed had been available (Figure 2U).

Shrubs

Two seeds, Atriplex nummularia and Sida rohlenae, were designated as being from shrubs but the distinctions between shrubs and trees is sometimes not clear (Beard 1981). As discussed later, the results for the two shrubs were ambiguous and did not produce clear indications of their utility as seeds for grinding. Further investigation of this group is required. Trees

Tree seeds in general showed great variation in their grinding performance (Figure 4), both in raw form and when pre-processed but, on occasion, approached the grass seeds in grinding productivity. However, in untreated form, only Acacia cowleana achieved this distinction (28 seconds). It is a hard, but brisant, seed and fractured readily under moderate muller pressure after which it ground into a fine, plastic meal (Figure 1P). The product had a slightly gritty feel but readily formed into cakes.

A dramatic example of the changes possible with pre-processing tree seeds was available from Acacia acuminata (raspberry jam wattle). In raw form, this seed could not be ground successfully: in 75 seconds, it produced only a rough meal which adhered to the muller and impeded grinding (Figure 1H & Figure 1I). However, when parched, it became the most productive tree seed to grind and, at 14 seconds, was the fifth fastest seed overall.

Experimental results for Acacia coriacea (wirewood) and Acacia victoriae (elegant wattle) did not fully accord with the ethnographic literature (e.g. Devitt 1992). A. coriacea seed cases are both hard and resilient (rubbery) and in raw form, were found to be virtually impossible to grind to any degree of fineness (116 seconds) (Figure 1N). For Acacia victoriae, all conditions resulted in slow times; the fastest being 59 seconds after parching. For some unknown reason, this seed did not grind well in small lots; however, with heavier loads it could be ground into a moderately coarse meal with sufficient adherence to be formed into cakes (Figure 1R).

The remaining tree seeds, Acacia aneura and Brachychiton acuminatus, could both be satisfactorily ground (in mid-range times) in clean, parched or soaked condition although B. acuminatus had a tendency to accumulate grinding product under the muller (Figure 2D).

96 Cane (1984; 1989) observed both sedges and succulents (herbs) and grasses being ground. While the seeds of both groups were small and soft, he noticed that the portulacas were the ‘easiest to grind of all the seeds in the region’ (1984:84). It appeared that he had discerned that differences in grinding times existed even between small, soft seeds. In our experiments, both groups of seeds were easily ground, and no difficulties were apparent. Grinding notes regularly commented on the ease with which the herbs could be ground but did not, at that early stage of the experiments, pinpoint the group as being noticeably easier to process than the grasses. However, a possible difference between such apparently similar seeds suggested further investigation was required.

Accordingly, an independent samples t-test was conducted to compare grinding times for the fast and moderate true native grasses (N = 19, M = 20.74, SD = 10.8) and the herbs [N = 12, M = 11.33, SD = 6.3; t (31) = 2.733, p = .011, two-tailed; Eta ² =.21]. The difference between the means of the two groups is statistically significant at the p = .05 level and the effect is considered large thus supporting the suggestion from Cane’s work that all soft seeds do not have the same grinding characteristics.

The native grasses (as above) were also compared with the fast and moderate tree seeds by a further independent samples t-test [N = 38, M = 32.03, SD = 11.7; t(57) = 3.533, p = .001, two-tailed; Eta² = .19]. The difference is statistically significant at the p = .05 level and the effect is large supporting the perceived differences in grinding effort between grass and tree seeds. (Grasses and shrubs were not compared due to problems with the shrub group).

It has been proposed that all Acacias and most tree seeds (exceptions being Brachychiton, Acacia victoriae and A. murrayana, ‘which must be cracked before they can be ground’) had a similar grinding time to A. aneura (O’Connell and Hawkes 1981; O’Connell et al. 1983:92). Accordingly, a comparison was made between mulga and the other four Acacias included in the experiments. After excluding those seed-condition combinations where grinding time per gram exceeded 55 seconds, 10 cases with mulga (M = 33.5, SD = 9.9) and 21 with the others (M = 31.1, SD = 13.3) were available. An independent samples t-test was conducted but was unable to establish any statistical difference between the two groups [t(31) = 0.526, p = .60, two-tailed] thus supporting O’Connell and Hawkes’ assumption that grinding times for Acacias generally may be similar to that of mulga.

However, when the five Acacias are compared to each other, individual variations became apparent. A One-Way between-groups ANOVA was conducted for the five seeds: F (4, 26) = 5.61, p = .002. The effect size, calculated using eta squared, was 0.46 suggesting a large effect. Post-hoc comparisons using the Tukey HSD test indicated that the mean score for Acacia acuminata (M = 18.50, SD = 6.32) was significantly different to Acacia aneura (M = 33.5, SD = 9.9), (un-cracked) Acacia victoriae (M = 40.0, SD = 6.16) and Acacia coriacea (M = 39.5, SD = 11.27). Comparison with the remaining A. cowleana (M = 28.2, SD =10.68) was not statistically significant. Acacia acuminata required only around half the time to grind as the other three — a real, as well as statistical, difference.

Effects of Seed Condition and Pre-Treatment on Processing Times

Due to excessive time required to grind or other problems with grinding, not all seed and treatment combinations were found to improve the grinding qualities of the seeds. Table 6 shows the results of processing times given different seed conditions.

With a range of treatments accessible to Aboriginal seed-users, it is likely that, in normal circumstances, only the most efficient combinations would have been adopted especially as both parching and soaking could be undertaken in bulk and time per gram was thus negligible. The seeds were predominantly ground in clean or raw condition. However, for the

97 native seeds where pre-treatment was utilised, the various seed-condition combinations are shown in Figure 5. The most effective (fastest to grind) combination for all seeds tested has been graphed in Figure 6.

Grinding Times by Seed Condition

75 A. acuminata 14 39 A. aneura 28 34 116 A. coriacea 54 68 Clean 64 Raw A. victoriae

SeedType 59 69 Parched 38 B. acuminatus 34 Soaked 36 15 37 S. rohlenae 20 39 0 20 40 60 80 100 120 140 Time in Seconds

Figure 5 - Grinding times in seconds by grinding condition for the pre-processed native seeds.

Figure 6 - Most efficient (fastest to grind) combination for each seed (the fast time for S. rohlenae is considered anomalous).

All the native grasses were soft but brisant under the muller (except the pliant panicum) and all reduced to fine flour with a minimum of effort. They generally ground equally well wet or dry and readily formed into cakes. In view of these favourable characteristics, no attempts were made to condition them by parching or soaking.

The seeds of the shrub Atriplex nummularia (old man saltbush) are extremely difficult to remove from their surrounding fruit and various attempts to do so resulted in the destruction of the caryopsis. Attempts to grind the fruit whole, both raw and parched, achieved only minor success with a product not unlike a rolled oats breakfast cereal rather than flour being slowly produced (Figure 2B). All grindings with this seed recorded slow times and have effectively been excluded from analyses. As no practical solution could be devised to lower the times or improve the grinding characteristics, further attempts to process A. nummularia to flour were abandoned.

Experimental results obtained for the second shrub, Sida rohlenae (shrub sida) also require heavy qualification. With this seed it was not so much the grinding time that was critical but rather the output. Grinding the spongy flower cases whole (raw) required 37 seconds and resulted not in flour being produced, but rather a very coarse meal (Figure 2N). Soaking had little effect (39 seconds) but parching reduced the grinding time by about half to 20 seconds

99 and placed the seed in the fast category. However, parching did nothing to improve the output quality. De-husking to a ‘clean’ seed by flash-burning the flowers substantially reduced the casing. This resulted in very fast grinding times (15 seconds) but charred the seeds to such an extent that they were likely inedible. In view of the difficulty in obtaining a satisfactory product or clean seed, we consider the experimental results significantly overestimated the likely real utility of this seed.

The tree seeds Acacia aneura (mulga) and Brachychiton acuminatus (kurrajong) were only modestly improved by pre-treatments. Both A. aneura (Figure 1K) and B. acuminatus performed best when the seeds were parched but only with mulga would the difference — 11 seconds— be likely to matter in practical terms.

While difficult to grind raw (116 seconds), parching Acacia coriacea (wirewood) seeds enabled them to be roughly ground and formed into cakes in 54 seconds but, as some whole or partially ground seeds usually remained, not to the consistency of flour or meal but more resembling a nut bar confectionery. The seeds appeared to grind better dry as wet grinding protected unground seeds from engaging the grinding surfaces. Even with prior parching, the black seed coat fragments clogged the muller surface and were difficult to remove. It is noted that Devitt reported the seeds being soaked to remove the arils and make a drink before being dried for further processing (1988:141). Experimental soaking for four hours did not remove the arils but improved the grinding time for the seed to, a still slow, 68 seconds.

For Acacia victoriae, experimental grinding did not support all the complex ethnographic pre- treatment procedures suggested (e.g. Kimber and Smith 1992), but confirmed that the usual pre-treatments alone were not sufficient for easy grinding. The clean seed is very hard but brittle and capable of being cracked by the muller of a millstone / muller pair although care needs to be taken to ensure seeds do not jump off the grinding surface. Soaked seeds required longer grinding times and is consistent with subjective impressions that soaking actually toughened the seeds. Parching provided a modest improvement (59 seconds) plus the seed became firm (moderately hard) and could be readily fractured by the muller. It proved to be a mid-range seed in terms of ease of grind and formed a plastic mass which readily moulded into cakes.

An attempt was made to pre-crack Acacia victoriae with a shallow mortar and a pebble pestle. However, with the grain samples and mortar available, the seed proved to be intractable. When the pestle was ‘pushed’ onto the seeds, little impression was made; when the pestle was pounded into the seeds, the seeds jumped out of the mortar and were lost in such quantities that the experiment was abandoned. The shallow mortar proved much less efficient for cracking these seeds than the millstone and muller used for grinding. The experimental results do not accord well with the wide-spread use of this seed. Additional testing of complex pre-treatments is needed to ascertain whether some combination exists which will reduce grinding times more in line with other popular seeds.

Bearing in mind the earlier comments on individual tree seeds, the effects of pre-treatments on this group were particularly apparent with seeds moving up or down in rank due to parching or soaking. For example, parched A. acuminata (14 seconds) headed this group and could compete with most grasses. Parched A. aneura (28 seconds) also performed well. However, while A. aneura was widely popular, grinding still required nearly three times as long as the fastest grasses or herbs (in addition to the minor inconvenience of parching).

All other aspects being equal, economies using the fast grass-type seeds as staples, such as Tindale’s Panara grass seed culture (1977), clearly possessed an inherent advantage over those economies having to utilise even the most accommodating of tree seeds.

100 Pre-Cracking of Seeds A number of hard seeds were said to require pre-processing by cracking before grinding, usually in a mortar (e.g. Smith 1985, 1989; Pardoe 2003). Preliminary experiments with four of the hard seeds available indicated that worthwhile time advantages may be available from separating the processing into pre-cracking and grinding. However, this must be viewed in context. For example, as mentioned earlier, with the popular A. victoriae, pre-cracking with the shallow, but widely available, mortar was not a successful experimental option and additional technology, such as the less common deep mortar, may have been necessary if pre-cracking was to be adopted. With domestic Sorghum bicolor, pre-cracking produced only a modest three second advantage. However, with the native seeds, widely used Acacia aneura and also Brachychiton acuminatus, pre-cracking and grinding clean seeds required only roughly half the time as did grinding the uncracked clean seeds — 20 and 21 seconds compared to 39 and 38 seconds. When the comparison was to parched seeds, the advantage reduced to eight seconds and 17 seconds, a still worthwhile benefit. Figure 7 illustrates the three successfully cracked clean seeds. Pre-cracking the two native seeds captured substantial economic gains and further demonstrated the economic value of the widely distributed shallow mortar.

Comparison of Grinding Conditions 45

35 Grinding Cracked 30 Cracking Grinding Uncracked 25

20 39 38 15 14 Grindingtime in seconds 7 17 10 19

5 9 6 4 0 Sorghum Acacia aneura Brachychiton bicolor acuminatus

Figure 7 - Average grinding time per gram for grinding un-cracked clean seeds and for cracking the seeds plus grinding the cracked grain.

However, if the advantage demonstrated with this small sample of seeds applies more widely and if other commonly utilised hard seeds cannot, like A. victoriae, be successfully cracked in a shallow mortar, it raises the question of why deep mortars do not appear more regularly in the archaeological record. One possible answer is that wooden, rather than stone, mortars and pestles were used (e.g. Meehan 1989; Steward 1933) but have not survived. This issue requires further investigation.

101 Effects of Seed Hardness on Processing Times

There is a general perception in the ethnographic literature that soft seeds are faster to grind than hard seeds (e.g. Cane 1989, O’Connell et al. 1983); that is, there is a relationship between seed hardness and grinding time. Our impressions when undertaking the grinding suggested that the very soft seeds, codes Hard 5 and Hard 4, were both similarly quick and easy to grind and that there was also little difference in grinding time between H2 and H3 coded seeds. It was only with the obviously very hard H1 seeds like A. victoriae and A. coriacea that noticeable extra time was required for grinding. H5 seeds were largely herbs and very small grasses and H4 seeds the remaining grasses and soft, conditioned Acacias. H3 seeds were predominantly parched tree seeds and H2 seeds mostly the remaining clean and raw Acacias plus white French millet (due to the pressure required in its initial cracking).

In order to ascertain whether seed hardness actually affected grinding times for fast and medium seed combinations (excluding the single Sorghum plumosum case), a one-way between groups analysis of variance was conducted. There is a statistically significant difference at the p < .05 level in grinding time scores for the five Hardness groups: (ASNOVA, F (4,102) = 4.2, p = .004). The effect size, calculated using eta squared, is large (0.14). Post-hoc comparisons using the Tukey HSD test indicate that the mean scores for groups H5 (M = 19.1, SD = 11.6) and H4 (M = 18.1, SD 7.7) are significantly different from that of group H1 (M = 31.6, SD = 13.9). Groups H2 and H3 do not vary significantly from the other groups. The approximately six second increase in average grinding time from H5 and H4 seeds to H2 and H3 seeds and the further increase of another six seconds to the hard H1 seeds is thus practically and statistically significant and there appears to be strong relationship between seed hardness and grinding time.

We can also compare Ease of Grinding and Fineness of Grind against Grinding Time. As Ease of Grinding and Grinding Time measured essentially similar factors, not surprisingly, the relationship was more or less constant. However, the relationship between Grinding Time and Fineness of Grind was not clear cut.

Fouling Residues

With some seed grinding, a waxy, reflective residue lodging deeply into the interstices of the stone on both upper and lower grindstones became a problem. This coating appears visually similar to ‘sickle gloss’ but is caused not by the reduction of the surficial grains in the sandstone matrix, but by a build-up of plant residue (see Vardi et al. 2010 for a discussion and references). These deposits inhibit further grinding due to their masking effect on the abrasive grinding surfaces and can be very difficult to remove. Removal may require rejuvenation of the grinding surface by pecking which in turn reduces the use-life of the grindstone.

When grinding the herbs and grasses, only minor fouling residue was noted: specifically with the five-gram load of A. lappacea and with T. pungens where the fouling residue appeared to be associated with the presence of residual husks on the seeds. For some other seed types, fouling residues may be of greater significance for grinding productivity. For example, with B. acuminatus, deposition of fouling was frequently observed and may have been accentuated when the seeds were pre-cracked. However, as partially ground seeds and meal constantly and firmly adhered to all grinding surfaces, accurate assessment of the extent of fouling residue from this seed was difficult. With mulga, previous experimental grindings of clean seeds (sample one) were impeded by the heavy deposition of residue on the surfaces of both millstone and muller (Figure 1L). For this current experiment, using seed from that same lot but some twelve months later (plus a new sample two), no fouling residue was generally noted. The exception was a minor accumulation when the seed was soaked

102

Proxies

Many Australian native seeds are difficult and expensive to obtain, especially in a condition suitable for grinding. As such, a preliminary assessment was made to ascertain whether any of the three domestic seeds— Panicum maximum (green panic) Panicum miliaceum, (white French millet) and Sorghum bicolor (red sorghum) — would prove to be viable grinding analogues for native seeds for experimental purposes. While morphology, in itself, seems to have little direct impact on how readily a seed can be ground, it was considered prudent to only compare seeds with some morphological similarities in this preliminary assessment.

No native seeds of similar morphology to Panicum miliaceum were able to be obtained and its usefulness as a proxy cannot be assessed until supplies of the larger native seeds like Brachiaria miliiformis (arm grass millet) can be obtained.

Sorghum bicolor is comparable in size to a number of the Acacias obtained for grinding; namely, A. acuminata, A. aneura and A. victoriae. An independent samples t-test was conducted to compare the scores for the sorghum group (M = 19.3, SD = 9.7) and Acacias group (M = 30.3, SD = 11.5; t (24) = -.211, p = .045, two-tailed). Although p was only slightly below .05, an eta squared value of .157 suggests a large effect. Contrary to our initial expectations, sorghum cannot be considered a general proxy for the Acacias. However, when the comparison was of sorghum to the two Acacias which had parched grinding times of less than 55 seconds (A. aneura and A. acuminata), a further t-test suggests that there is no significant difference between these groups (p = .558). This small sample (N= 13) indicated that sorghum may provide a useful proxy for the common pre-treated conditions of various Acacias.

Panicum maximum (green panic) was unable to be distinguished from the native Panicum decompositum (native millet) in the experiments. In addition, an independent-samples t-test could not establish any statistically significant difference between the two seeds, [t (12) = .302, p = .77, two-tailed]. Panicum maximum is thus a readily available ‘perfect’ proxy which could be substituted for native millet in experimental situations.

The comparison was then extended to include not only P. decompositum but also A. lappacea, S. rohlenae and T. pungens — plants with similar seed dimensions. A second t- test (p = .711) revealed no statistical difference between P. maximum and the group. Consequently, P. maximum should be able to be employed with confidence as a proxy for grass-type seeds with similar morphology.

A third comparison was made with the very small seeded plants, the lovegrass E. eriopoda and the herbs P. oleracea and T. verrucosa. An independent samples t-test between these two groups, P. maximum (M = 21.2, SD = 7.5) and the very small seeds (M = 11.8, SD = 6.3), reveals a statistical difference between the means [(t (22) = 3.04, p = .006, two tailed)]. The magnitude of the difference is large (eta squared = .295). Accordingly, P. maximum should not be used as an analogue for the very small seeds plants. For these, similarly small domestic seeds such as teff, amaranth and finger millet will need to be investigated.

The use of proxies is considered a potentially rewarding method of investigating the grinding qualities of difficult to obtain or very expensive seeds. However, our work to date can only be considered exploratory and further investigations are planned.

103 Discussion

This study constitutes the only comprehensive comparison of native seed grinding attributes and output rates yet undertaken for Australian plant species. Used as quantitative building blocks in conjunction with other data, for example, standardised nutrition tables (e.g. Brand- Miller and Cherikoff 1985) or use wear analyses (e.g. Hayes et al. 2018), the results should help archaeologists test questions surrounding past nutrition, diet breadth, species selection and seasonal plant contributions. The data should also illuminate aspects of wider issues like technological functionality, residential patterns and labour contributions to name a few examples. However, it must be kept in mind that the data are experimentally derived and are subject to the usual limitations of experimental archaeology (see, for example, Clarkson and Shipton 2015 and Coles 1979). In particular, extrapolating these brief times to the sometimes quoted ethnographic rate of ‘hours to grind a kilogram of seed’ may well introduce significant error.

The intent and expectation of these experiments was threefold. Firstly, to explore whether there were significant differences in the times required to grind various traditionally used native seeds as Mildwaters and Clarkson (2018) found with domestic seeds. Secondly, to investigate whether various methods of pre-treating the seeds (parching, soaking or cracking) materially altered their grinding times. Finally, to ascertain if domesticates can be substituted as viable proxies for difficult to obtain native seeds in investigating the grinding properties of the native seeds.

We believe that the first two questions have been positively answered; processing times depend on the type and condition of the seed being ground.

As detailed individually, substantive differences were established between various categories of native seeds. Even with ‘soft’ seeds, differences were noted — grasses varied from fast to medium and herbs were statistically faster to grind than the grasses. The tree seed group, including the Acacias, varied widely. For example, reasons for the extensive use of A. coriacea seed ethnographically are not evident from the experiments. Although the past importance of the seed may to some extent depend on its use when green, sufficient reports of its consumption when fully dry suggest that further investigation is required. Pre-cracking at some stage of preparation, probably after parching, would be a reasonable possibility if the grinding rate assumed by O’Connell and Hawkes (1981) was to be met. As three of the five experimental Acacias showed meaningful differences, assumptions of uniformity within the over 40 Acacias known to be ground should not be accepted unless confirmed by wider experimentation.

Pre-processing was of importance mainly with the tree seeds and revealed mixed results. Pre-cracking, in some instances, may have gained worthwhile time benefits but perhaps only at an increased investment in technology through a bulky and heavy deep mortar. Regardless, some seeds generally thought to require pre-cracking were shown to be satisfactorily processed with a typical millstone and muller pair. Soaking softened some seeds and thus reduced grinding times. However, one instance was encountered where soaking appeared to harden the seed. Parching was a worthwhile benefit in most cases, especially as it could be performed in bulk and, like soaking, required little time investment.

Individual seeds and seeds as groups clearly display wide production differences which need to be accounted for in any detailed assessment of traditional lifeways. The benefits associated with easily collected and ground soft seeds have been illustrated. Negative differences include any factors likely to incur penalties in processing time or reduce the use- life of the grindstones. Problems with de-husking and the deposition of fouling residue have

104 been briefly mentioned but are unlikely to be exhaustive. Clearly, it is misleading to speak of economies being based simply on grass seeds or tree seeds — the specific seeds being used must be known and fully evaluated. For the seeds examined, the experimental data will assist with more fine-grained modelling. For example, in conjunction with plant fruiting schedules, construction of time specific seasonal production profiles for individual seeds in an economy should be possible.

The third question is not conducive to a simple answer. The domestic seeds produced mixed results as proxies. Panicum miliaceum could not be assessed due to an absence of comparable native seeds. Sorghum bicolor should not be utilised as a general proxy for any of the untreated Acacia seeds tested. However, while insufficient cases were available to test the relationships between sorghum and various Acacias in their usual pre-treated use conditions, a small sample of parched Acacias showed no statistical difference to the sorghum suggesting that more in depth investigation is warranted. Employed as a proxy, Panicum maximum or green panic, could not be differentiated, either experimentally or statistically, from the widely exploited Panicum decompositum. Accordingly, its value in an experimental sense was established as it can be used as a direct analogue for native millet in experiments and also, with slightly less confidence, for many native seeds of similar morphology. However, considerable additional experimentation will be required to produce a comprehensive range of proxies for even the most commonly used native seeds.

The grinding qualities of seeds can be summarised as: if soft grass or herb seeds in pure form are involved, they will likely be problem free, fast grinding and not require pre- processing. Unlike the grasses, hard seeds from trees and shrubs display great variation, both in grinding ability and need for, or possible response to, the usually applied pre- treatments. Assessment of grinding characteristics for these seeds must be on an individual basis.

Conclusion

The paper has identified and recorded the grinding characteristics of a range of seeds and the data generated are modular, comparative and transferrable to a range of other situations. The experiments have demonstrated conclusively that the type and conditioning of the seed being ground are (along with grindstone functional surface area – see Clarkson and Shipton (2015) and Mildwaters and Clarkson (2018)) of major significance to seed processing times.

Under traditional circumstances, we find that the species of seeds used must have altered the labour committed to grinding flour by significant hours each day and thus a group’s capacity to address critical issues. The effects of these factors on food production in seed- based subsistence regimes are of such importance that major economic, technological and social decisions in Australia and elsewhere were likely influenced by seed type and availability.

However, the most rigorous experimental data is no substitute for the real-world know-how of Aboriginal and other indigenous peoples with intimate knowledge of seed use. It is critical that specific efforts are made to work with traditional owners in Australia and elsewhere to gather such knowledge for future generations and with which to assess the reliability of experiments such as these.

105 Acknowledgements

Special thanks are due to Dr. Ian Chivers of Native Seeds Pty. Ltd. and Nindethana Australian Native Seeds and to the two anonymous reviewers whose comments did much to improve the paper.

106 References

Allen, Harry. 1974. The Bagundji of the Darling Basin: Cereal Gatherers in an Uncertain Environment. World Archaeology. Vol. 5, No. 3. pp. 309-322. Anjun, F. M. and Walker, C. E. 1991. Review on the significance of starch and protein to wheat kernel hardness. Journal of the Science of Food and Agriculture. Vol. 56. pp. 1-13. Arndt, W. 1961. Indigenous sorghum as food and in myth: the Tagoman Tribe. Oceania. Vol. 32, No. 2. pp. 109-112. Beard, J. S. 1981. Vegetation of Central Australia. In John Jessup (Ed.) Flora of Central Australia. pp. xxi-xxvi. Sydney: Reed. Belderok B., Mesdag J., Donner D.A. 2000. Milling of wheat. In Donner D.A. (Ed.) Bread- making quality of wheat. Pp. 21-29. Springer, Dordrecht Bindon, Peter. 1996. Useful bush plants. Perth, WA: Western Australian Museum. Bird, Caroline and Beeck, Colin. 1988. Traditional plant foods in the southwest of Western Australia: the evidence from salvage ethnography. In Betty Meehan and Rhys Jones (Eds), Archaeology with Ethnography: An Australian perspective. pp. 113-122. Canberra: Australian National University. Bonney, Neville. 1994. What seed is that? A field guide to the identification, collection and germination of native seed in South Australia. Beverley, SA: N. Bonney. Brand-Miller, J. C. and Cherikoff, V. 1985. Australian Aboriginal bushfoods: the nutritional composition of plants from arid and semi-arid areas. Australian Aboriginal Studies. No 2. pp. 38-46) Bright, Jason, Ugan, Andrew and Hunsaker, Lori. 2002. The effect of handling time on subsistence technology. World Archaeology. Vol 34, No. 1. pp. 164-181. Cane, Scott. 1984. Desert Camps: a case study of stone artefacts and Aboriginal behaviour in the Western Desert. Unpublished Ph.D. thesis. Canberra: The Australian National University. Cane, Scott. 1987. Australian Aboriginal Subsistence in the Western Desert. Human Ecology. Vol. 15, No. 4. pp. 391-434. Cane, Scott. 1989. Australian Aboriginal Seed Grinding and its archaeological record: a case study from the Western desert. In David R. Harris and Gordon C. Hillman (Eds), Foraging and farming: the evolution of plant exploitation. pp. 99-119. Sydney: Unwin Hyman. Clarkson, Chris and Shipton, Ceri. 2015. Teaching ancient technology using “Hands-On” learning and experimental archaeology. Ethnoarchaeology. Vol. 7, No. 2. pp. 157- 172. Cleland, J. B. 1957. Our natives and the vegetation of southern Australia. Mankind, Vol. 5, No. 4. pp. 149-162. Cleland, J.B. 1966 The ecology of the Aboriginal in South and Central Australia. In B.C. Cotton (ed.), Aboriginal Man in South and Central Australia, pp.111-158. Adelaide: W. L. Hawes, Government Printer. Cleland, J. B. and Tindale, N. B. 1954. The ecological surroundings of the Ngalia natives in Central Australia and native names and uses of plants. Transactions of the Royal Society of South Australia. Vol. 77. pp. 81-86. Coles, John. 1979. Experimental Archaeology. London: Academic Press.

107 Devitt, J. 1988. Contemporary Aboriginal women and subsistence in remote, arid Australia. Unpublished Ph.D. thesis. Brisbane: University of Queensland. Devitt, Jeannie. 1992. Acacias: a traditional Aboriginal food source in central Australia. In A. P. N. House and C. E. Harwood (Eds) Australian dry-zone acacias for human food: proceedings of a workshop held at Glen Helen, Northern Territory, Australia, 7-10 August 1991. pp. 37-53.Melbourne: CSIRO Publications. Dillon, S. L., Lawrence, P. K., Henry, R. J., Price, H. J. and Johnstone, J. S. 2004. Sorghum laxiflorum and S. macrospermum, the Australian native species most closely related to the cultivated S. bicolor based on ITSI and ndhF sequence analysis of 25 Sorghum species. Plant Systematics and Evolution. Vol. 249. pp. 233-246. Edwards, D.A. and O’Connell, J. F. 1995. Broad spectrum diets in arid Australia. Antiquity. No. 69. pp. 769-83. Gason, Samuel. 1886. The Dieyerie Tribe of Australian Aborigines. In Edward M. Curr, The Australian Race: Its origin, languages, custom …Vol. 2. pp. 44-107. Melbourne: Government Printer. Goddard, Cliff and Kalotas, Arpad. 2002. Punu: Yankunytjatjara plant use. Alice Springs, N.T.: Jukurrpa Books. Golson, Jack. 1971. Australian Aboriginal Food Plants: Some Ecological and Culture- Historical Implications. In D. J. Mulvaney and J. Golson (Eds) Aboriginal Man and Environment in Australia. pp. 196-238. Canberra: Australian National University Press. Hayes, Elspeth, Pardoe, Colin and Fullagar, Richard. 2018.Sandstone grinding/ pounding tools: Use-trace reference libraries and Australian archaeological applications. Journal of Archaeological Science: Reports. Vol. 20, pp. 97-114. Horne, G. and Aiston, G. 1924. Savage life in Central Australia. London: Macmillan and Co. Isaacs, Jennifer. 1987. Bush food and herbal medicine. McMahons Point, N.S.W.: Weldons. Kimber, R. G. and Smith, M. A. 1987. An Aranda Ceremony. In D. J. Mulvaney and J. Peter White (Eds), Australians to 1788. pp.221. Broadway NSW: Fairfax, Syme & Weldon. Latz, P. K. 1995. Bushfires and Bushtucker: Aboriginal plant use in Central Australia. Alice Springs: IAD Press. Mallett, K. (Ed.). 2005. Flora of Australia Volume 44B. Melbourne: ABRS/CSIRO. Mallett, K. and Orchard, A., (Eds). 2002. Flora of Australia Volume 43. Melbourne: ABRS/CSIRO. Meehan, Betty. 1989. Plant Use in a Contemporary Aboriginal Community and Prehistoric Implications. Tempus. Vol. 1. pp. 14-30. Mildwaters, John. 2018. Seed-grinding stones: a review from a mainly Australian perspective. The Artefact 2016, Vol. 39. pp. 30-41. Mildwaters, John and Clarkson, Chris. 2018. The efficiency of Australian grindstones for processing seed: A quantitative experiment using reproduction implements and controlling for morphometric variation and grinding techniques. Journal of Archaeological Science: Reports. Vol. 17, pp. 7-18. Mitchell, T. L. 1839. Three expeditions into the interior of eastern Australia with descriptions of the recently explored region of Australia Felix . Vol. 1. London: T. & W. Boone. Mitchell, T. L. 1848. Journal of an Expedition into the Interior of Tropical Australia, in search of a route from Sydney to the Gulf of Carpentaria. London: Longman, Brown, Green, and Longmans. Newland, Simpson 1921. The annual address of the president. Proceedings of the Royal Geographical Society of Australasia (South Australian Branch). pp. 1-15.

108 O’Connell, J.F. and Hawkes, K. 1981. Alyawara plant use and optimal foraging theory. In B. Winterhalder and E. A. Smith (Eds.) Hunter-Gatherer Foraging Strategies: Ethnographic and Archaeological Analyses. pp. 99-125.Chicago: University of Chicago Press. O’Connell, J. F., Latz, P. K. and Barnett, P. 1983. Traditional and modern plant use among the Alyawarra of Central Australia. Economic Botany. Vol. 37, No. 1. pp. 80-109. Pardoe, Colin. 2003. The Menindee Lakes: a regional archaeology. Australian Archaeology. No. 57. pp. 42-53. Petersen, Nicolas. 1978. The traditional pattern of subsistence to 1975. In B. S. Hetzel and H. J. Frith (Eds), The nutrition of Aborigines in relation to the ecosystem of Central Australia. pp. 25-35. Melbourne: CSIRO Rao, K. S. 1991. Characteristics and fatty acid composition of Brachychiton species seeds and oils (Sterculiaceae). Journal of Agricultural and Food Chemistry. Vol. 39. pp. 881-882. Roth, W. E. 1897. Ethnological studies among the North-West-Central Queensland Aborigines. Brisbane: Govt. Printer. Roth, W. E. 1901. Food: Its Search, Capture, and Preparation. North Queensland Ethnography. Bulletin No. 3. Simms, Steven R. 1987. Behavioral Ecology and Hunter-Gatherer Foraging: An example from the Great Basin. Oxford: BAR International Series 381. Smith, M. A. 1985. A morphological comparison of Central Australian seedgrinding implements and Australian Pleistocene-age grindstones. The Beagle, Occasional Papers of the Northern Territory Museum of Arts and Sciences. Vol. 2, No. 1. pp. 23- 38. Smith, M. A. 1986. The antiquity of seedgrinding in arid Australia. Archaeology in Oceania. Vol. 21, No. 1. pp. 29-39. Smith, M. A. 1989. Seed gathering in inland Australia: current evidence from seed-grinders on the antiquity of the ethnohistorical pattern of exploitation. In David R. Harris and Gordon C. Hillman (Eds), Foraging and farming: the evolution of plant exploitation. pp. 305-317. Sydney: Unwin Hyman. Smith, M. A. 2004. The grindstone assemblage from Puritjarra rock shelter: investigating the history of seed-based economies in arid Australia. In Tim Murray (Ed.), Archaeology from Australia, pp. 168-186. Melbourne: Australian Scholarly Publishing. Smith, M. A. 2013. The archaeology of Australia’s deserts. Cambridge: Cambridge University Press. Spencer, Baldwin and Gillen, F. J. 1899. The Native Tribes of Central Australia. New York: Dover. Stenvert, N. L. and Kingswood, K. 1977. The influence of the physical structure of the protein matrix on wheat hardness. Journal of the Science of Food and Agriculture. Vol. 28. pp. 11-19. Steward, Julian H. 1933. Ethnography of the Owens Valley Paiute. University of California Publications in American Archaeology and Ethnology. Vol. 33, No. 3. pp. 238-249. Sturt, Charles. 1849. Narrative of an expedition into central Australia, performed under the authority of Her Majesty's Government, during the years 1844, 5, and 6: together with a notice of the province of South Australia, in 1847. Vol.1. London: T. and W. Boone. Sweedman, Luke. 2006. Australian seeds: a photographic guide. In Luke Sweedman and David Merritt (Eds), Australian seeds: a guide to their collection, identification and biology. pp. 67-172. Collingwood, Vic.: CSIRO Publishing.

109 Tindale, N. B. 1972. The Pitjandjara. In M. G. Bicchieri (Ed.) Hunters and Gatherers Today: A Socioeconomic Study of Eleven Such Cultures in the Twentieth Century. pp. 217- 265. New York: Holt Rinehart and Winston. Tindale, N. B. 1977. Adaptive significance of the panara or grass seed culture of Australia. In R. V. S. Wright (Ed.) Stone tools as cultural markers: change, evolution and complexity. pp. 345-349. Canberra: A.I.A.S. Turner, Fred. 1895. Australian Grasses. Sydney: Government Printer. Vardi, Jacob, Golan, Amir, Levy, Daniel and Gilead, Isaac.2010. Tracing sickle blade levels of wear and discard patterns: a new sickle gloss quantification method. Journal of Archaeological Science. Vol. 37, pp. 1716-1724.

110

5 Chapter 5 - Summary and Conclusions

5.1 Summary

Australian ethnographic seed-grinding information is archaeologically important and has been utilised by a number of researchers, within and outside Australia, in developing, for example, theory at various ranges. In particular, the performance characteristics of ground seed use—and the grindstone technology supporting it— were identified as being especially significant in the drier areas of the world, including Australia, with its extensive grass and shrub lands. However, both the existing ethnographic data on seed-grinding and the knowledge-base about fundamental grinding technology display omissions or problems when closely scrutinised. The three papers forming the core of this thesis examined a number of these problems and were able to clarify various aspects through controlled experiments.

As discussed more fully in Chapter 2, the ethnographic data reveal a number of possibly serious methodological issues when extrapolated beyond their initial contexts. For example, only 20 instances detailing quantified Australian ethnographic grinding production (of flour or meal) were able to be located in the literature— a modest corpus for the problems they have been expected to explicate. However, even the 20 instances were not all autonomous and free-standing. The data were reported by six research teams but only four principal researchers were involved — Brokensha (1975), Cane (1984), O’Connell (and Hawkes 1981) and Smith (1986; 1989; 2013; 2015; et al. 2015). Fourteen instances involved the actual observation of grinding and six were estimates based on the reports of others— five of the six utilising the work of Brokensha. However, Brokensha himself did not report a specific rate for grinding flour but rather the amount of cooked damper produced. Further, one of the actual instances was likely the same experiment counted twice. It will be apparent that all the 20 instances are not independent observations: rather, many involve circular relationships. Only the six observations by Cane are clearly actual and independent.

111

In view of this paucity, it is important that researchers understand the true limitations of the available data.

Chapters 3 and 4 addressed a number of specific issues about the seed-grinding puzzle including the effects on grinding performance of millstone size, grinding action, muller configuration and seed variety. The grindstone size (more specifically, the functional grinding surface area) and the type of seed were the most significant factors determining performance.

Large millstones out-performed the smaller millstones and grindstones but no critical size threshold was apparent at around 300 mm as suggested by Smith (1985:26). Grinding efficiency increased in a more or less linear progression with the functional surface area of the grindstones. Very small grindstones have been identified as seed-grinding platforms (Veth and O’Connor 1996:21) and the experiments confirmed that small grindstones are capable of grinding seeds at useful volumes. However, unless exceptional circumstances prevailed, their low efficiency would be unlikely to make them a preferred choice for large-scale intensive processing of seeds. The large millstones were the most efficient and perhaps warranted their usual designation as formal, specialised or ‘distinctive’ ‘seedgrinders’ (Smith 1986:38). However, as discussed above, the real problem arises in deciding which implements are not ‘seedgrinders’; dichotomous classifications of Australian grindstones appear to lead to as many problems as solutions (for example, see discussions of Smith’s typology by Gorecki et al. (1997) and Veth and O’Connor (1996)). All the ‘new’ grindstones evidenced at least slight formation wear in the form of incipient groove development at the end of the experiments so, on a strict application of Smith’s principles, could be said to have changed from amorphous grindstones to formal seed-grinders — a paradoxical result for normal use. Further, all the smaller grindstones despite (like Pleistocene and Early Holocene archaeological examples) not resembling Smith’s ‘distinctive implements’ were found to have worthwhile seed-grinding capability. Their grinding capability appears such that seed consumption as a supplement should have been possible from early times (Clarkson et al. 2017) and not needed to await the introduction of specialised millstone technology in the middle to late Holocene (Smith 1986). Seed use could

112

thus have been both earlier and more widespread than presently proposed (Gorecki et al. 1997).

The second major factor in grinding productivity identified was the type of seed being ground. Surprisingly, this is an aspect seldom addressed by researchers, the two most notable exceptions being Cane (1984) and O’Connell and Hawkes (1981; et al. 1983). The seeds being processed, and pre-grinding treatments such as parching and soaking, were major determinants in the varying production of flour or meal. The three proxy domestic seeds and the native seeds all had differing grinding profiles. An average time of around four hours for the production of a kilogram of wet meal has often been adopted by researchers as a rule-of-thumb following the seminal work of O’Connell and Hawkes (1981). However, in the experiments, different combinations of grindstone and seed type changed the time required to produce a kilogram of meal over a very wide range of approximately one to eight hours. Differences in labour investment of such magnitudes from the accepted ‘average’ were likely to have influenced whether seeds could be viably incorporated into an economy. Accordingly, reassessment of seed rankings in formal foraging models (e.g. O’Connell et al. 1983) may be advisable.

Six seeds— described as fast— were found to be able to be ground in 22 seconds or less, on average, for one gram of seed. As would be expected, these included two of the most widely utilised grasses, Eragrostis eriopoda (Woollybutt) and Panicum decompositum (native millet). Two succulents / herbs, Portulaca oleracea (pigweed) and Tecticornia verrucosa (samphire) were also included in the fast category. Whilst pigweed was known from the ethnographic literature to be easy to grind, the inclusion of T. verrucosa was unexpected. Also unexpected was the inclusion of a tree seed, Acacia cowleana or Halls Creek wattle, which is not often mentioned in the literature. None of these five seeds required pre-treatment. The sixth fast seed, Brachychiton acuminatus, a variety of kurrajong or bottle tree, was soaked before grinding. Many soft grass-type seeds were difficult to obtain in quantities suitable for experimentation and, since they all ground without difficulty, were not considered for pre-treatment.

113

Of the pre-treatments (mainly of tree seeds) prior cracking of the grain in a mortar shortened grinding time (sometimes significantly) but the cracking constituted such a high proportion of the overall processing that it needed to be treated as a separate procedure and thus was excluded from the experiments. Contra Smith (1985:27) it was noted that some very hard seeds (in particular Acacia victoriae used by Smith as an example of a seed requiring mortar cracking) could not always be successfully cracked in the common, widely distributed, traditional shallow mortar. As such, if any benefits of pre-cracking were to be realised, extension of the traditional toolkit by the incorporation of a deep mortar may have been necessary. Experiments designed to investigate these aspects would be beneficial. However, it should be noted that some seeds usually thought to require treatment in a mortar were found to be successfully processed on the millstone without pre-cracking (again including A. victoriae).

The other pre-treatments, parching and soaking, could be performed in bulk in a traditional setting and thus their costs to overall processing times can be considered nominal. Results were mixed but, in general, parching provided a worthwhile benefit with hard seeds.

In summary, where soft, clean grass or herb seeds are available, they will likely prove to be fast and easy to grind. While some herbs will possibly be slightly faster to process, the difference should not be noticeable in practical circumstances. However, with the hard tree or shrub seeds, wide variation is likely to be encountered in both grinding time and response to pre-treatments. Each seed needs to be individually assessed. As proposed by Tindale for the Panara Grass-seed Culture (1977), economies able to utilise ‘grass-type’ seeds had a significant labour investment advantage for the grinding component of processing over those using ‘tree’ seeds.

Many native seeds were expensive and only able to be obtained in small quantities. Accordingly, further extension of these experiments is advocated as different seeds become available, especially in quantities sufficient to allow grinding for ten-minute grinding sessions as were utilised in testing the grindstones in Chapter 3. These data would enable estimates to be made of how much seed could be ground (or wet meal

114

produced) in an hour – the measurements favoured in the literature– and thus allow direct comparison with much of the ethnographic data.

The general perception of seed-grinding is that a diagnostic wet grinding technique was normally utilised (e.g. Smith 1985; Tindale 1977). In reality, in traditional circumstances both dry and wet methods were used, sometimes in combination, according to need (e.g. Roth 1897; Horne and Aiston 1924). The diagnostic aspect of wet grinding appears to relate to wear occasioned when ground meal was swept off the distal end of the millstone (Smith 1985, Fig. 3d). Why wet meal would cause wear, and dry meal would not, is not clear. Experiments with both wet and dry grinding were conducted and no diagnostic differences were perceived in the relatively brief use-wear period encompassed by the experiments. However, it was established that grain throughput was about 60% greater with wet grinding; a significant advantage for the wet method.

The type of grinding action utilised, reciprocal or rotary, was found to be of minor importance with millstones of more-or-less equal functional area having similar performance. However, this raised questions requiring investigation about the utility of dish or basin millstones which were common in the south-easterly grass areas of dry Australia. The functional grinding area of a basin millstone used with a rotary action is much greater than that of a grooved millstone of equal length but requires proportionally greater raw material in its creation. If performance is comparable, why would sometimes scarce stone be utilised in this way? Experiments with the functional grinding area of both the grooved and basin millstones kept constant are needed to clarify this issue. (It is worth mentioning that analogous issues arise in the Americas, where specific metates (basal grindstones) of differing morphology but comparable functional areas, are considered essential under a range of specialist conditions).

Top stones or mullers were not specifically investigated although a number of observations were possible. The large dished or basin millstones discussed above were able to be operated with a wide variety of mullers whereas the very small or deeply grooved grindstones were sensitive to the selection of a topstone. It is

115

suggested that both grinding action and muller technology would both benefit from further targeted experimental investigation.

With the passage of time, fewer and fewer Aboriginal people who ground seeds for a living will be available to be interviewed. However with, for example, elders on outstations endeavouring to hand down knowledge and maintain aspects of ‘the old ways’, it is possible that a few people from various regions will still possess some knowledge of seed use and grinding and may still be able to demonstrate valuable skills to archaeologists. It is impossible to predict outcomes but interviews with even indirectly knowledgeable Aboriginal informants directed specifically towards grindstones would be of immense value. While quotidian enquiries would need to be a priority and widely directed — from acquisition to disposal for grindstones and from identification to consumption for seeds — other social, technical and economic factors such as reasons for the inclusion or rejection of certain seeds from a diet as well as preferences based on taste or ease of gathering/handling, would be equally important. Realistically, any one informant may not be in possession of a full range of grinding knowledge, however each piece of information may allow another important piece of the jigsaw to be fitted into the whole. Ethnoarchaeology was developed to provide ethnographic data useful to archaeologists and analysis of modern ethnoarchaeological reports (for example those mentioned in Table I of the Introduction) illustrates how relevant and content laden such research can be even when compiled from fragmentary information. Every effort should be made to obtain any ethnographic seed-grinding knowledge relevant to archaeology that is still capable of being collected.

In the absence of ethnoarchaeological data, it is considered that experimental research such as that forming the core of this thesis is the best available vehicle to answer some of the basic questions which still confuse our views of grindstones and seed-grinding. Controlled experiments are capable of generating quantifiable and transportable fresh data that can reliably inform new research.

In addition to the uses made of seed-grinding data in model construction, few other aspects of our present understanding of grindstones (and their technology) are problem free. This includes the term ‘grindstone’ itself, as grindstone typologies are

116

confused and out of date. In the absence of a comprehensive, generally accepted typology, each use of a term by a researcher requires detailed individual definition if confusion is to (at least) be minimised — a number of aspects of the ‘Smith-Gorecki Debate’ turn on quite fine definitional distinctions (Gorecki et al. 1997; Smith 1985, 1986; Veth and O’Connor 1996). In essence, the fundamentals of seed-grinding are still not fully understood and documented — in Adams’ (2014) terms, the basic technological issues have not been adequately resolved.

In earlier chapters and above, a number of pieces of the seed-grinding jigsaw currently generating debates were identified. It was suggested that experimental investigation in various areas would be beneficial and the experiments reported here have provided a measure of clarity concerning several of these issues. However, the matters discussed barely touch the surface of the basic technological issues which need to be resolved. Further concerns which would benefit from experimental investigation include- a. Assessment of a wider range of grinding implements. Those investigated in Chapter 2 could largely be described as ‘central Australian type seed- grinders’ (Smith 1985). However, other tools had the potential to contribute to a seed-using economy. Such implements as the deep mortars mentioned above, tjiwa grindstones (Smith 2015), dimpled grindstones (Bird and Beeck 1988), and axe-grinding grooves (Pardoe 2015), warrant examination as do implements made of wood. b. Investigation of variations in performance due to the raw material used in the grindstones. In the Americas, the raw materials from which grindstones were created were considered of major importance (Adams 1999; 2002). Whilst sandstone, and to a lesser extent, quartzite predominated in Australia, other rocks were sometimes used and any effects produced from different raw materials need to be known. c. Examination of the effects of maintenance on grindstone performance. Many grindstones required maintenance by periodically roughening (pecking) the grinding surfaces and the consequences of this for performance and the use- life of the implement should be known. Also knowing whether altering the texture of the grinding surface could substitute for the use of alternative raw materials would be of interest.

117

d. Analysis of the contribution of the top-stone to grinding productivity. Whilst a start has been made in determining performance differences between bottom stones, a full investigation of the variation attributable to a range of top stones is necessary (Hard et al. 1996). e. Development of experimental accelerated wear models. The morphology of both bottom and top grindstones and their wear profiles has attracted regular debate and some suggestions as to how they derive their ever changing forms and the import of such factors have been proposed. (e. g. Gorecki et al. 1997; Smith 1985; 1986). The morphology at any one point in time may be the results of a combination of numerous factors including the raw materials from which they were created, the grinding actions used, the maintenance regime adopted, what was ground, whether it was ground wet or dry, the initial configuration of the slab and so on. The relatively slow wear rates of the stones make investigation of these aspects difficult. It is suggested that experiments with softer rocks (or possibly controlled hardness artificial stones) which could accelerate wear, may yield useful information on the relative contributions of various factors to wear regimes

At 65ka, Australia now has the oldest known plant grinding outside of Africa (Clarkson et al. 2017) and, unlike most research in other parts of the world, grinding technology in Australia is not influenced by the processing of farmed domesticated plants. As such, Australia provides a unique platform to investigate exclusively hunter- gatherer seed-grinding at great time depth.

5.2 Concluding Remarks

It is believed that this thesis and the three core papers have fulfilled their intended objectives by clarifying some important pieces of the grinding puzzle. They provide researchers with a number of basic, but clearly defined, building blocks which can be used as foundations for the future; a future that can now be addressed with a little less uncertainty and without the necessity of extensive and cumbersome qualifications concerning the aspects assessed.

118

By identifying areas which could fruitfully be further investigated using experimental treatments, investigators may be encouraged to revisit, and extend, the various seed-grinding debates; perhaps identifying and incorporating areas and technologies at present not normally associated with Aboriginal seed-grinding. Dobres stresses that the ‘traditional orientation employed to study prehistoric technology is oriented to the macroscale’ (Dobres 1995:28). Finer-grained assessments are needed and it is hoped that this work will encourage further examination at a range of scales.

119

5.3 References

For ‘Introduction’ and ‘Conclusions’ only.

Adams, Jenny L. 1993. Toward understanding the technological development of manos and metates. Kiva. Vol. 58, No.3. pp. 331-344.

Adams, Jenny L. 1999. Refocusing the role of food-grinding tools as correlates for subsistence strategies in the U.S. Southwest. American Antiquity. Vol. 64, No.3. pp. 475-498.

Adams, Jenny L. 2002. Ground Stone Analysis: A Technological Approach. Salt Lake City: University of Utah Press.

Adams, Jenny L. 2010. Understanding Grinding Technology through Experimentation. In Jeffery R. Ferguson (Ed.) Designing Experimental Research in Archaeology: Examining Technology Through Production. pp. 129-151. Boulder: University Press of Colorado.

Adams, Jenny L. 2014. Ground Stone Analysis: A Technological Approach, Second Edition. Salt Lake City: University of Utah Press.

Adams, Jenny, Delgado, Selina, Dubreuil, Laure, Hamon, Caroline, Plisson, Hugues and Risch, Roberto. 2009. Functional analysis of macro-lithic artefacts: a focus on working surfaces. In F. Sternke, L. Eigeland and L-J Costa (Eds) Non-Flint Raw Material Use in Prehistory: Old prejudices and new directions. pp. 43-66. Oxford: BAR International Series 1939.Akerman, Kim, Fullagar, Richard and Van Gijn, Annelou. 2002. Weapons and wunan: production, function and exchange of Kimberley points. Australian Aboriginal Studies. Vol.1. p.p. 13-42.

Allen, Harry. 1974. The Bagundji of the Darling Basin: Cereal Gatherers in an Uncertain Environment. World Archaeology. Vol. 5, No. 3. pp. 309-322.

Balme, Jane. 1991. The antiquity of grinding stones in semi-arid Western New South Wales. Australian Archaeology. No. 32. pp. 2-9.

Bartlett, Katharine. 1933. Pueblo milling stones of the Flagstaff Region and their relation to others in the Southwest. Museum of Northern Arizona Bulletin. No.3. pp. 5-32.

Bentley, R. Alexander, Lipo, Carl, Maschner, Herbert D. G. and Marler, Ben. 2008. Darwinian Archaeologies. In Bentley, R. Alexander, Maschner, Herbert D. G. and Chippindale, Christopher (Eds), Handbook of archaeological theories. Lanham: Alta Mira Press. pp. 124-141.

Beveridge, Peter. 1889. The Aborigines of Victoria and Riverina as seen by Peter Beveridge. Melbourne: M. L. Hutchinson.

Bird, Caroline and Beeck, Colin. 1988. Traditional plant Foods in the Southwest of Western Australia: the evidence from salvage ethnography. In Betty Meehan

120

and Rhys Jones (Eds) Archaeology with Ethnography: An Australian perspective. pp. 113-122. Canberra: Australian National University.

Bird, Douglas W. and O’Connell, James F. 2006. Behavioral ecology and archaeology. Journal of Archaeological Research. Vol. 14. pp. 143-188.

Bright, Jason, Ugan, Andrew and Hunsaker, Lori. 2002. The effect of handling time on subsistence technology. World Archaeology. Vol 34, No. 1. pp. 164-181.

Brokensha, Peter. 1975. The Pitjantjatjara and their Crafts. North Sydney: Aboriginal Arts Board.

Bulmer, John. 1887. Some account of the Aborigines of the Lower Murray, Wimmera, Gippsland and Maneroo. Proceedings of the Royal geographical Society of Australasia, Victorian Branch. Vol. 5, Part 1. pp. 15-43.

Cane, Scott. 1984. Desert Camps: a case study of stone artefacts and Aboriginal behaviour in the Western Desert. Unpublished Ph.D. thesis. Canberra: The Australian National University

Clarkson, Chris, 2007. Lithics in the land of the lightning brothers: the archaeology of Wardaman Country, Northern Territory. Acton: ANU E Press.

Clarkson, Chris., Jacobs, Zenobia., Marwick, Ben., Fullagar, Richard., Wallis, Lynley., Smith, Mike., Roberts, Richard G., Hayes, Elspeth., Lowe, Kelsey., Carah, Xavier., Florin, S.Anna., McNeil, Jessica., Cox, Delyth., Arnold, Lee J., Hua, Quan., Huntley, Jillian., Brand, Helen E. A., Manne, Tiina., Fairbain, Andrew., Shulmeister, James., Lyle, Lindsay., Salinas, Makiah., Page, Mara., Connell, Kate., Park, Gayoung., Norman, Kasih., Murphy, Tessa and Pardoe, Colin. 2017. Human occupation of northern Australia by 65,000 years ago. Nature. 547 pp. 306-310.

Clarkson, Chris., Smith, Mike., Marwick, Ben., Fullagar, Richard., Wallis, Lynley A., Faulkner, Patrick., Manne, Tiina., Hayes, Elspeth., Roberts, Richard G., Jacobs, Zenobia., Carah, Xavier., Lowe, Kelsey M., Matthews, Jacqueline and Florin, S. Anna. 2015. The archaeology, chronology and stratigraphy of Madjedbebe (Malakunanja II): a site in northern Australia with early occupation. Journal of Human Evolution. Vol. 83. pp. 46-64.

Clarkson, Chris and Shipton, Ceri. 2015. Teaching ancient technology using “hands- on” learning and experimental archaeology. Ethnoarchaeology. Vol. 7, No. 2. pp. 157-172.

Cleland, J. B. and Johnston, T. H. 1937. Notes on native names and uses of plants in the Musgrave Ranges region [Part 1]. Oceania. Vol. 8, No. 2. pp. 208-215.

Coles, John. 1973. Archaeology by Experiment. London: Routledge.

Coles, John. 1979. Experimental Archaeology. London: Academic Press.

121

David, Nicholas. 1998. The Ethnoarchaeology and Field Archaeology of grinding at Sukur, Adamawa State, Nigeria. African Archaeological review. Vol. 15, No.1.pp. 13-63.

Delgado-Raack, Selina and Risch, Roberto. 2009. Towards a systematic analysis of grain processing technologies. In M. Araujo Igreja and C. Conte (Eds), Recent Functional Studies on Non Flint Stone Tools: Methodological Improvements and Archaeological Inferences, 23-25 May 2008 Lisboa. Padrao des Descobrimentos: Lisbon. pp. 1-20.

Delgado-Raack, Selina and Risch, Roberto. 2016. Bronze Age cereal processing in Southern Iberia: A material approach to the production and use of grinding equipment. Journal of Lithic Studies. Vol.3, No. 3. pp. 125-145

Dobres, Marcia-Anne. 1995. Gender and pre-historic technology: on the social agency of technical strategies. World Archaeology. Vol. 27, No.1. pp. 25-49.

Dodson, John, Fullagar, Richard and Head, Lesley. 1992. Dynamics of environment and people in the forested crescents of temperate Australia. In John Dodson (Ed.), The Naïve Lands: Prehistory and environmental change in Australia and the south-west Pacific. pp. 115-159. Melbourne: Longman Cheshire.

Dubreuil, Laure. 2001. Functional studies of prehistoric grindstones: a methodological research. Bulletin du Centre de recherché francais a Jerusalem. Vol. 9. pp. 73-87.

Edwards, D.A. and O’Connell, J. F. 1995. Broad spectrum diets in arid Australia. Antiquity. No. 69. pp. 769-83.

Eitam, D., Kislev, M., Karty, A. and Bar-Yosef, O. 2015. Experimental barley-flour production in 12,500-year-old roc-cut mortars in Southwestern Asia. PLoS One 10(7): e0133306. https://doi.org/10.1371/journal.pone.0133306

Field, Judith and Fullagar, Richard. 1998. Grinding and pounding stones from Cuddie Springs and Jinmium. In Richard Fullagar (Ed.) A Closer Look: Recent Australian Studies of Stone Tools. pp. 96-108. Sydney: Sydney University Archaeological Methods Series.

Florin, S. Anna and Carah, Xavier. 2018. Moving past the ‘Neolithic problem’: the development and interaction of subsistence systems across northern Sahul. Quaternary International. Vol. 489.pp. 46-62.

Fullagar, Richard. 1994. Traces of times past: stone artefacts into prehistory. Australian Archaeology. No. 39. pp. 63-73.

Fullagar, Richard. 2006. Starch grains, stone tools and modern hominin behaviour. In Sean Ulm and Ian Lilley (Eds), An Archaeological Life: Papers in honour of Jay Hall. pp.191-200. Brisbane: University of Queensland.

Fullagar, Richard & Field, Judith. 1997. Pleistocene seed-grinding implements from the Australian arid zone. Antiquity. Vol.71, No. 272. pp. 300-307.

122

Gorecki, Paul and Grant, Miranda. 1994. Grinding patches from the Croydon region, Gulf of Carpentaria. In M. O’Sullivan, S. Brockwell and A. Webb (Eds) Archaeology in the north: proceedings of the 1993 Australian Archaeological Association Conference. pp. 227-246. Darwin: North Australian research Unit.

Gorecki, P., Grant, M., O’Connor, S. and Veth, P. 1997. The morphology, function and antiquity of Australian grinding implements. Archaeology in Oceania. Vol. 32, No. 2. pp. 141-150.

Haaland, Randi. 1995. Sedentism, cultivation, and plant domestication in the Holocene Middle Nile Region. Journal of Field Archaeology, Vol. 22. No. 2. pp.157-174.

Hamilton, Annette. 1980. Dual Social Systems: Technology, Labour and Women’s Secret Rites in the eastern Western Desert of Australia. Oceania. Vol. 51, No. 1. pp. 4-19.

Hamon, Caroline. 2008. Functional analysis of stone grinding and polishing tools from the earliest Neolithic of north-western Europe. Journal of Archaeological Science. Vol. 35. pp. 1502-1520.

Hard, Robert J., Mauldin, Raymond P. and Raymond, Gerry R. 1996. Mano size, stable carbon isotope ratios, and macrobotanical remains as multiple lines of evidence of maize dependence in the American Southwest. Journal of Archaeological Method and Theory. Vol. 3, No. 4. pp. 253-318.

Hayden, Brian. 1987. Traditional metate manufacturing in Guatemala using chipped stone tools. In Brian Hayden (Ed.) Lithic studies among the contemporary highland Maya. pp. 8-119. Tucson: The University of Arizona Press.

Hayes. Elspeth. 2015. What was ground? A functional analysis of grinding stones from Madjedbebe and Lake Mungo, Australia. Unpublished Ph.D. thesis. Wollongong: University of Wollongong.

Hiscock, Peter. 2001. Sizing up prehistory: sample size and composition of artefact assemblages. Australian Aboriginal Studies. Vol. 1. pp. 48-62

Hiscock, Peter. 2017. Discovery curves, colonisation and Madjedbebe. Australian Archaeology. Vol. 83, No. 3. pp. 168-171.

Hiscock, Peter and Kershaw, A. Peter. 1992. Palaeoenvironments and prehistory of Australia’s tropical Top End. In John Dodson (Ed.), The Naïve Lands: Prehistory and environmental change in Australia and the south-west Pacific. pp. 43-75. Melbourne: Longman Cheshire.

Horne, G. and Aiston, G. 1924. Savage life in Central Australia. London: Macmillan and Co.

Horsfall, Gayel A. 1987. Design theory and grinding stones. In Brian Hayden (Ed.) Lithic studies among the contemporary highland Maya. pp. 332-377. Tucson: The University of Arizona Press.Howitt, A. W. The Aborigines of Cooper’s

123

Creek. In R. Brough Smyth, The Aborigines of Victoria: with notes relating to the habits of the natives of other parts of Australia and Tasmania compiled from various sources for the Government of Victoria. Vol.II. pp. 305. Melbourne: Govt. Printer.

Huffman, Thomas N. 2006. Maize grindstones, Madikwe pottery and ochre mining in precolonial South Africa. Southern African Humanities. Vol. ,No. 2. pp. 51-70.

Kraybill, Nancy. 1977. Pre-Agricultural Tools for the Preparation of Foods in the Old World. In Charles A. Reed (Ed.) Origins of Agriculture. pp. 485-521. The Hague: Mouton Publishers.Kroeber, A. L. 1925. Handbook of the Indians of California. Bureau of American Ethnology Bulletin 78. Washington: Government Printing Office.

Lidstrom Holmberg, Cecilia L. 1998. Prehistoric Grinding Tools as Metaphorical Traces of the Past. Current Swedish Archaeology. Vol.6. pp. 123-142.

Lourandos, Harry. 1997. Continent of Hunter-Gatherers: New Perspectives in Australian Prehistory. Cambridge: Cambridge University Press.

Loy, T. H. 1992. Blood residue analysis of 90,000-year-old stone tools from Tabun Cave, Israel. Antiquity. Vol. 66. Pp. 24-36.

McCarthy, F. D. 1976. Australian Aboriginal Stone Implements. Sydney: Australian Museum.

McNiven, Ian J. 1995. Fissioning and regionalisation: the social dimensions of changes in Aboriginal use of the Great Sandy Region, southeast Queensland. In Jay Hall and Ian J. McNiven (Eds), Australian Coastal Archaeology. pp. 157-168. Canberra: Australian National University.

Madsen, David B. and Simms, Steven R. 1998. The Fremont complex: a behavioral perspective. Journal of World Prehistory. Vol. 12, No. 3. pp. 255-336.

Marchant, Leslie R. 1988. An Island Unto Itself: William Dampier and New Holland. Carlisle WA: Hesperian.

Mathieu, James R. 2002. Introduction, in James R. Mathieu (Ed.), Experimental Archaeology: Replicating past objects, behaviors, and processes. pp. 1-11. Oxford: BAR International Series 1035.

Mathieu, James R. and Meyer, Daniel A. 2002. Reconceptualizing experimental archaeology: assessing the process of experimentation, in James R. Mathieu (Ed.), Experimental Archaeology: Replicating past objects, behaviors, and processes. pp.73-82. Oxford: BAR International Series 1035.

Mauldin, Raymond. 1993. The Relationship Between Ground Stone and Agricultural Intensification in Western New Mexico. Kiva. Vol.58, No. 3. pp. 317-330.

124

Mildwaters, John. 2005. Economic and social consumption in the Aboriginal material world: ‘More was going on …’. Unpublished Honours Thesis. University of Queensland, School of Social Science, Brisbane.

Mitchell, T. L. 1848. Journal of an Expedition into the Interior of Tropical Australia, in search of a route from Sydney to the Gulf of Carpentaria. London: Longman, Brown, Green, and Longmans.

Nixon-Darcus, L. and A. C. D’Andrea. 2017. Necessary for life: studies of ancient and modern grinding stones in highland Ethiopia. African Archaeological Review. Vol. 34. pp. 193-223.

O’Connell, J.F. and Hawkes, K. 1981. Alyawara plant use and optimal foraging theory. In B. Winterhalder and E. A. Smith (Eds.) Hunter-Gatherer Foraging Strategies: Ethnographic and Archaeological Analyses. pp. 99-125.Chicago: University of Chicago Press.

O’Connell, J. F., Latz, P. K. and Barnett, P. 1983. Traditional and modern plant use among the Alyawarra of Central Australia. Economic Botany. Vol. 37, No. 1. pp. 80-109.

O'Connor, S. and Chappell J. 2003. Colonisation and coastal subsistence in Australia and Papua New Guinea: different timing, different modes. In C. Sand (Ed.) Pacific Archaeology: Assessments and Prospects. Proceedings of the International Conference for the 50th Anniversary of the First Lapita Excavation (July 1952). pp. 15-32. Noumea: Departement d'Archaeologie, Service des Musses et du Patrimoine.

Outram, Alan K. 2008. Introduction to experimental archaeology. World Archaeology. Vol. 40, No. 1. pp. 1-6.

Pardoe, Colin. 1994. Bioscapes: The evolutionary landscape of Australia. Archaeology in Oceania. Vol. 29, No. 3. pp. 182-190.

Pardoe, Colin. 2015. Grinding stones of the Lachlan River; archaeological studies along the Lachlan River; Condobolin, Lake Cargelligo, West Wyalong and Hillston regions. Part of the research project ’Kiacatoo Man: biology, archaeology and environment at the Last Glacial Maximum. March 2015.

Pastron, Allen G. 1974. Preliminary ethnoarchaeological investigations among the Tarahumara. In C. B. Donnan and C. W. Clewloe (Eds), Ethnoarchaeology. pp. 93-116. Los Angeles: University of California.

Piperno, Dolores R., Weiss, Ehud, Holst, Irene and Nadel, Dani. 2004. Processing of wild cereal grains in the Upper Palaeolithic revealed by starch grain analysis. Nature. Vol. 430. pp. 670-673.

125

Robitaille, Jerome, 2016. The ground stone industry of the Mursi of Maki, Ethiopia: Ethnoarchaeological research on milling and crushing equipment (technique and function). Journal of Lithic Studies. Vol. 3, No. 3. pp. 429-456.

Roth, W. E. 1897. Ethnological studies among the North-West-Central Queensland Aborigines. Brisbane: Govt. Printer.

Rowan, Yorke M. and Ebeling, Jennie R. 2008. Introduction: The potential of ground stone studies. In Y. M. Rowan and J. R. Ebeling (Eds), New Approaches to Old Stones: Recent studies of ground stone artifacts. pp. 1-15. London: Equinox.

Rowley-Conway, Peter and Layton, Robert. 2011. Foraging and farming as niche construction: stable and unstable adaptations. Philosophical Transactions of the Royal Society. Vol. 386. pp. 849-862.

Schiffer, Michael B., Skibo, James M., Boelke, Tamara C., Neupert, Mark A. and Aronson, Meridith.1994. Perspectives on experimental archaeology: surface treatments and thermal response of the clay cooking pot. American Antiquity. Vol. 59, No. 2. pp. 197-217.

Schneider, Joan S. 1996. Quarrying and production of milling implements at Antelope Hill, Arizona. Journal of Field Archaeology. Vol. 23, No. 3. pp. 299- 311.

Searcy, Michael T. 2011. The Life-Giving Stone. Tucson: University of Arizona Press.

Semeniuk, V. 1995. The Holocene record of climatic, eustatic and tectonic events along the coastal zone of Western Australia: a review. Journal of Coastal Research, Special Issue No. 17. pp. 247-259.

Shanks, Michael and Tilley, Christopher. 1987. Re-Constructing Archaeology: Theory and Practice. Cambridge: Cambridge University Press.

Shoemaker, Anna C., Davies, Matthew I. J. and Moore, Henrietta L. 2017. Back to the grindstone? The archaeological potential of grinding-stone studies in Africa with reference to contemporary grinding practices in Marakwet, Northwest Kenya. African Archaeological Review. Vol. 34. pp. 415-435.

Simms, Steven R. 1985. Acquisition costs and nutritional data on Great Basin resources. Journal of California and Great Basin Anthropology. Vol. 7, No. 1. pp. 117-126.

Simms, Steven R., O’Connell, James F, and Jones, Kevin T. 2013. Some thoughts on evolution, ecology, and archaeology in the Great basin. In Parezo, Nancy J. and Janetski, Joel C. (Eds), Archaeology in the Great Basin and Southwest, pp. 177-188.Salt Lake City: University of Utah Press.

Smith, M. A. 1985. A morphological comparison of Central Australian seedgrinding implements and Australian Pleistocene-age grindstones. The Beagle,

126

Occasional Papers of the Northern Territory Museum of Arts and Sciences. Vol. 2, No. 1. pp. 23-38.

Smith, M. A. 1986. The antiquity of seedgrinding in arid Australia. Archaeology in Oceania. Vol. 21, No. 1. pp. 29-39.

Smith, M. A, 1988. Central Australian seed grinding implements and Pleistocene-age grindstones. In B. Meehan and R. Jones (eds) Archaeology with Ethnography: an Australian Perspective, pp. 94-108. Canberra: The Australian National University.

Smith, M. A. 1989. Seed gathering in inland Australia: current evidence from seed- grinders on the antiquity of the ethnohistorical pattern of exploitation. In David R. Harris and Gordon C. Hillman (Eds), Foraging and farming: the evolution of plant exploitation. pp. 305-317. Sydney: Unwin Hyman.

Smith, M. A. 2013. The archaeology of Australia’s deserts. Cambridge: Cambridge University Press.

Smith, Mike. 2015. Western Desert tjiwa-and tjungari- type grindstones and their archaeological significance. Australian Aboriginal Studies. Vol. 1. pp. 115-121.

Smith, Mike, Hayes, Elspeth and Stephenson, Birgitta. 2015. Mapping a millstone: the dynamics of use-wear and residues on a Central Australian seed-grinding implement. Australian Archaeology. Vol. 80. pp. 70-79.

Spencer, Baldwin and Gillen, F. J. 1899. The Native Tribes of Central Australia. London: Macmillan.

Stahl, A. B. 1989. Plant-food processing: implications for dietary quality. In David R. Harris and Gordon C. Hillman (Eds), Foraging and farming: the evolution of plant exploitation. pp. 171-194. Sydney: Unwin Hyman.

Stern, Nicola., Tumney, Jacqueline, Fitzsimmons, Kathryn E. and Kajewski, Paul. 2013. Strategies for investigating human responses to changes in landscape and climate at Lake Mungo in the Willandra Lakes, southeast Australia. In David Frankel, Susan Lawrence, and Jennifer Webb (Eds) Archaeology in Environment and Technology : Intersections and Transformations. pp. 31-50. New York: Routledge. Steward, Julian H. 1933. Ethnography of the Owens Valley Paiute. University of California Publications in American Archaeology and Ethnology. Vol. 33, No. 3. pp. 238-249.

Stone, Tammy. 1994. The impact of raw material scarcity on ground-stone manufacture and use: an example from the Phoenix Basin Hohokam. American Antiquity. Vol. 59, No. 4. pp. 680-694.

Tindale, N. B. 1977. Adaptive significance of the panara or grass seed culture of Australia. In R. V. S. Wright (Ed.) Stone tools as cultural markers: change, evolution and complexity. pp. 345-349. Canberra: A.I.A.S

127

Van Peer, Philip, Fullagar, R., Stokes, Steven, Bailey, R. M. , Moeyersons, Jan, Steenhoudt, Frans, Geerts, A., Vanderbeken, T., De Dapper, Morgan and Geus. F. 2003. The early to middle Stone Age transition and the emergence of Modern Human behaviour at site 8-B-11, Sai Island, Sudan. Journal of Human Evolution. Vol. 45. pp. 187-193.

Veth, Peter, Fullagar, Richard and Gould, Richard. 1997. Residue and use-wear analysis of grinding implements from Puntutjarpa Rockshelter in the Western Desert: Current and proposed research. Australian Archaeology. No. 44. pp. 23-25.

Veth, Peter and O’Connor, Sue. 1996. A preliminary analysis of basal grindstones from the Carnarvon Range, Little Sandy Desert. Australian Archaeology. No. 43. pp. 20-22.

Veth, Peter, Susan O’Connor and Lynley A. Wallis. 2000. Perspectives on ecological approached in Australian archaeology. Australian Archaeology. No. 50. pp. 54-65.

Vinnicombe, Patricia. 1987. Salvage archaeology of the Burrup Peninsular. Australian Archaeology. No. 25. pp. 53-79.

Willcox, George and Stordeur, Danielle. 2012. Large-scale cereal processing before domestication during the tenth millennium cal BC in northern Syria. Antiquity. Vol. 86, No. 331. pp. 99-114.

Winterhalder Bruce and Smith, Eric Alden. 2017. Evolutionary ecology and the social sciences. In E. A. Smith and B. Winterhalder (Eds), Evolutionary Ecology and Human Behavior. pp. 3-23. New York: Routledge.

Wright, Mona K. 1993. Simulated use of experimental maize grinding tools from southwestern Colorado. Kiva. Vol. 58, No. 3. pp. 345-355.

128